PRE2017 4 Groep5

Group Members

Name	Study	Student ID
Ahmed Ahres	Software Science	0978238
Quinten Maes	Psychology & Technology	0955972
Hugo Melchers	Mathematics & Software Science	0994280
Christel van den Nieuwenhuizen	Psychology & Technology	0940672
Frank de Veld	Physics & Mathematics	1010914

Introduction

Automation is one of the biggest changes taking place in industry today. Nowadays, systems are being automated to the extent that they require almost no human intervention. Such a technology has been successful not only in manufacturing, but also in the automotive industries. Self-driving cars or even drones are examples of where automation has seen research and development in flight.

Suitcases are also an example that can benefit from automation. By making suitcases autonomously follow their owner, we can facilitate the traveling process by making it less tiring, especially for the elderly, disabled people and pregnant women. Moreover, this would allow a more efficient transportation of clothes and objects, which could be useful for business travelers or regular flyers.The user group grows every year if we look at the increasing amount of flying travelers, see fig 1. Research has shown that people were enthusiastic and trustworthy towards a prototype of a robotic suitcase after using it . However, one must keep in mind aspects such as security, object detection and components to be able to fit in a plane according to airport security standards.

We first planned to develop a smart self-following suitcase in order to make traveling and transporting objects more efficient and less tiring. This system required different technologies: Bluetooth connection to be able to follow a device owned by a person, GPS tracking to avoid getting lost by the owner, security procedures to avoid getting stolen and computer vision for obstacle avoidance. The plan was to make framework to build such a system, present an object detection algorithm that can be used by the suitcase for obstacle avoidance as well as a user interface for a mobile application that can be used by the owner of our system. However, after deliberation it became clear that this was a very complex project and that an easier solution exists for the same problem, namely an engine assisting suitcase which relieves the load of the suitcase by detecting the direction of movement from the pull of the user and using a small motor to ride in that direction itself too. There is thus a shared-control system present. With a separate smartphone application it is possible to tune the amount of work done by the suitcase. The advantages of this idea is that there are no trust-issues from users since the users maintain control over the suitcase as well as physical contact. The risk of theft is also reduced and the environmental detection system is removed which annuls privacy issues. Furthermore, without object detection and obstacle avoidance complex hardware and software architecture are not needed which also reduces the computational power required and increases battery life.

Problem Statement

The first proposed problem statement was: How can we securely make traveling and transporting objects more efficient through autonomous suitcases?

The smart system created should follow its owner within a certain of range of distance, be protected from theft and avoid getting lost.

However, the idea was switched to an engine-assisting suitcase and the problem statement needed to be updated to the following:

How can we make traveling and transporting objects more efficient through engine-assisting suitcases, keeping user's preferences in mind? Since this proposed idea and technology is easier to produce and analyze, it should also be applicable and useful in different environments than airports.

Project Planning

Approach

The original approach for the self-following suitcase was the following:

We aim to design a motorised suitcase that can be configured to follow its owner. First, we will develop a vision for a nominal use case of such a motorised suitcase. From this vision, we will extract user requirements. Then, we will give a high-level architecture showing the different components that will be used (e.g. computer vision, electric drive, etc.). We will do research regarding all these components, gathering the knowledge required to combine them into one device. Finally, we will present a design for a motorised suitcase based on our vision, the user requirements, and the technicalities of the used components used. To deliver an actual working prototype is not the aim, as it turns out that even the companies working on such suitcases (see State of the Art) are having major problems with that, and making such a prototype would require a budget of several hundred euro's if not thousands. Furthermore, the time available for this project seems too little to build a complete prototype and we probably lack the expertise to combine all the needed systems together. Thus, we only aim at designing the prototype and the hardware/software systems for such a prototype as well as the consequences of such a technology. As the total technology architecture of a self-following suitcase is complex in itself, we expect that the time needed for this project and the time available for this project will match well enough.

However, since the idea was adapted to a engine-assisting suitcase, the approach was slightly changed. Most notably the vision system disappears completely from the suitcase and the control system becomes much easier. However, additionally, different sensors were needed for measuring the work done by the user for pulling the suitcase and the computer in the suitcase should autonomously process this input and convert it to the necessary output. Since this new idea is in theory easier than the self-following suitcase, more attention should be given to the accompanying smartphone application as well as a more careful study for the hardware and the movement system. The planning was adapted apropriately.

Weekly Planning

Revisions

Our weekly planning was changed and adapted over time, some points were moved around due to time constraints or because of a change in work division. One of the biggest revisions to the planning was made when we decided together with the tutor to go more in depth on some of the specifications as they were to global as well as agreeing on potentially performing some experiments using the soccer robots named TURTLES, from the Tech United Student team from the university of Eindhoven. We would slightly modify them in terms of appearance and coding to more closely mimic the behaviour of a suitcase.

Some other ideas that came to light during the revisions were the elaboration of the mobile application that supports the functionalities of the suitcase as well as the idea to make a functioning prototype which has (some of) the functions that we want to implement in the final product.

Sadly, it quickly became apparent that we would not be able to use the TURTLES as these robots are currently being transported to Canada where they will compete in the World Championship. Because of this we decided to put more effort into getting a working smartphone application.

Milestones

The following milestones are set for this course. They are shown in the planning aswell.

• In the first week every group member finds and summarizes the articles for their part for the state of the art. The entire introduction and accompanying parts are written.
• In the second week the literature study is done.
• In the third week the requirements and UI design are finished.
• In the fourth week the USE analysis is finished.
• In the fifth week an in-depth analysis of the hardware system is completed
• In the eighth week the mobile application is finished for a first prototype.
• The last milestone is the completion of all deliverables.

Deliverables

For our end deliverables we have to do the following:

• A presentation, which will be held in the last week. In this presentation, we will discuss our findings, present a solution to the problem, and, if possible, give a demonstration.
• This wiki page containing the technology mentioned earlier, taking into account the advantages, disadvantages, costs, and impact of such an implementation. This wiki page also contains an in-depth study of the engine-assisting mechanism as well as a user interface for the desired mobile application, making all of it a framework to develop such a system. Furthermore, a mobile application will be made for this project which implements the user interface which will be made as well.

Users

In general the users of this technology are people who regularly have to cover distances with baggage and either have trouble with that or would like to have their load relieved. This mostly covers business travelers, pregnant women, elderly and disabled people. The general area of use would thus be airports, train stations, bus stations, naval docks and urban areas. It is preferable that these technologies are used in relatively smooth and flat terrain. In the case of personal suitcases, the engine-assisting suitcases would be bought particularly by the users. Sections 13.1 and 13.2 detail more about the user needs and user impact.

Requirements

ID	Category	Requirement
R1	Airline regulations	The battery for the suitcase shall be removable.
R2		The weight of the suitcase shall meet the relevant airline regulations, which means it shall be significantly less than the maximum allowed weight.
R3		The battery for the suitcase shall meet airline regulations regarding material and maximum electric power.
R4		The size of the suitcase shall meet airline regulations regarding handluggage and hold baggage.
R5		The suitcase shall adhere to the TSA policies of the United States control border.
R6	Work reduction	The suitcase shall be able to reduce the amount of work done by the user by 50% at the minimum.
R7	Battery Life	The battery of the suitcase shall have a battery life of up to 10 kilometers of walking distance.
R8	Sensor	The suitcase shall be able to detect the amount of work done by the user with an uncertainty of 5% at maximum.
R9		The suitcase shall be able to detect the location of the user in a radius up to 50 meters via Bluetooth connection.
R10	Alarms	The suitcase shall transmit an urgent alarm to the user's phone when outside a radius of 50 meters of the user.
R11		The suitcase shall transmit an alarm when the battery life of the suitcase becomes lower than 10%.
R12		The smartphone application belonging to the suitcase system shall transmit an alarm when the battery life of the phone becomes lower than 10%.
R13	Mobile Application	The smartphone application belonging to the suitcase system shall be free to download.
R14		The smartphone application belonging to the suitcase system shall be available on the Android Store.
R15		The smartphone application belonging to the suitcase system shall be available on the Apple Store.
R16		The smartphone application belonging to the suitcase system shall display the weight of the suitcase.
R17		The smartphone application belonging to the suitcase system shall display the (remaining) battery life of the suitcase.
R18		The smartphone application belonging to the suitcase system shall display the remaining walking distance of the suitcase.
R19		The smartphone application belonging to the suitcase system shall lock and unlock the suitcase.
R20		The smartphone application belonging to the suitcase system shall (dis)connect from and to multiple suitcases.
R21		The smartphone application belonging to the suitcase system shall turn the suitcase on or off.
R22		The smartphone application belonging to the suitcase system shall configure the movement protocol of the suitcase.
R23	Portability	The suitcase shall have both an on mode and a manual mode in which the user pulls the suitcase himself without assisting.
R24	Location	The smartphone app belonging to the suitcase system shall allow the user to access the location of the suitcase.
R25	Costs	The cost of the final product shall not exceed 700 euros.
R26	Materials	The suitcase shall be made out of sturdy and shock-damping materials.
R27	Minimal Weight and Space	The components of the suitcase which are not in a regular suitcase shall be of minimal weight and occupy a minimal amount of space.
R28	Battery	The battery of the suitcase shall be rechargeable.

State of the Art

Self-following suitcase

So far, a few companies have already tackled the case of self following suitcases. At several events, such devices have been shown or demonstrated, but so far none have been sold to private individuals. The companies currently working on a self following suitcase are:

• Olive robotics owned by IKAP Robotics from Iran
• Travelmate robotics from the United States
• ForwardX from China
• Cowarobot owned by LeMoreLab from China
• 90FUN owned by Shanghai Runmi Technology Co., Ltd. from China
• NUA Robotics from Israel

It is important to note that all of these companies, with the parent companies included, have been founded in the previous 3 years. Two of these companies, namely Travelmate and Cowarobot, required crowdfunding from the crowdfunding website indiegogo.com in order to be able to produce their suitcases, Travelmate even needed two (). Both companies had delivery dates around end 2016 / begin 2017, but both experienced production delays and have not yet delivered the suitcases to their customers. Furthermore, Travelmate has doubled to tripled their prices compared to their original plan (€400 to €600 increases), which was comparable to the price of the one from Cowarobot. Nevertheless, both companies received an overwhelming amount of support from their backers; 400% of the necessary funds was received by Cowarobot and 1600% of the necessary funds were received by Travelmate. The other four companies do not give specifications on either the release date or the price, only NUA Robotics and ForwardX hope to release it at the end of 2018.

It is also interesting to look at the specifications of all these self-following suitcases. Due to airline regulations, most of the companies have a removable battery for the suitcase or aim at having that implemented. Not all companies give specifications on the battery, but most seem to have a Lithum-Ion battery. ForwardX also gives specifications on the capacity: 96 Wh, which meets IATA standards for carry-on luggage. For finding the owner of the suitcase, all companies either have a Bluetooth connection between the suitcase and the smartphone of the user or a Bluetooth connection between the suitcase and an additional smart wristband the user needs to wear. Some companies also use GPS or 3G/4G systems for location detection. The speed of the suitcases varies; for example NUA robotics is only capable of letting the suitcase go 5 km/h while 90FUN promotes a maximum speed of 18 km/h. All suitcases can ride at walking speed however. At last, the sensors for scanning the surroundings differ per company. NUA Robotics and Olive Robotics use a (stereoscopic) camera, ForwardX uses an ultrasonic sensor, Cowarobot a combination of both, while 90FUN and Travelmate don't appear to use an environment measuring sensor.

Moreover, five of these six companies claim to have made the first self following suitcase; only ForwardX is not claiming this. Additionally, NUA Robotics is a very small company and the company does not look very professional.

To summarize; there are several companies currently working on the idea of a self following suitcase and they are at varying stages of releasing the technology in form of a product, but it seems most of the companies are too optimistic or have run into some issues and none are delivering yet. However, since the companies are already promoting their idea, there are also numerous responses from potential buyers, both negative and positive. This is a very important source of feedback, since this technology is largely based on the users and their preferences. The most important positive remarks are:

• It is an innovative idea, meaning that certain groups of people will be interested in it in any case
• It is efficient
• It can help disabled people, elderly and pregnant women to transport their suitcases without health risks

The most important negative remarks or fears are:

• It could be stolen easily
• It could be vulnerable to hacking
• The weight could become too high, since the allowed weight for hand luggage is limited
• The size of the batteries might be too high, since the guidelines for batteries on aircraft are strict
• Terrorism could become easier
• It might not useful enough for the target group. This can of course be avoided by searching for other applications, such as aiding elderly or disabled people and improving efficiency in work environments.

It is important that these subjects are adequately thought of in the design of such a suitcase.

However, the case of self following trolleys does not seem to have been explored yet. One device which resembles this idea is the Stewart Golf X9 Follow , a self following golf cart. Another device is a prototype of a self following shopping cart . The golf cart seems to be the only one of the products listed in this section which is actually for sale. Thus, no self following trolleys have been made for airports, retail stores, train stations or other places where a lot of products need to be moved. Since the golf cart is the only actual product resembling the technologies aimed for in this product, it is interesting to investigate how this company has approached the problem of self-following trolleys. The golf cart uses Bluetooth and a separate remote which the user should keep with him and the range is 50 meters. The cart rides a few meters behind the user when on ‘Follow-mode’, but apparently it sometimes ‘chases rabbits’. The battery life is 25 to 30 holes, which means that it last a few hours long. Furthermore, it has downhill braking and an integrated stabilizer, which are important to have on golf courses. Of course, this machine is only useful for transporting golf clubs and golf accessories.

For the intended devices, several important (software) components and concepts are needed, of which there generally is a lot of research to be found. The important software concepts are autonomous image recognition, obstacle avoidance, remote tracking, control systems, autonomous driving and Bluetooth/GPS connections. Hardware subjects that are relevant are battery efficiency, the most optimal motors and wheels, Arduino connections and battery life. Furthermore, there are general subjects like privacy, security, user preferences and (airport) regulations.

Engine-assisting suitcase

Some of the information on the self-following suitcase is actually still useful when researching an engine-assisting suitcase. Most notably this includes most of the hardware and specifications, most of the applications for which the system is used (including the applications for self-following trolleys), the issues the companies working on the self-following suitcase encountered, and very importantly the feedback given by users on the idea of the self-following suitcase. As it turns out, most people are still afraid of either a malfunctioning robot or theft, both resulting in the loss of the luggage. Another frequent remark was that the self-following suitcase formed either a too complex solution to a problem or a solution for a problem that did not exist. Of course, disabled, pregnant or unfit people are then disregarded, but it is an important remark. The idea of an engine-assisting suitcase largely solves these important issues, as well as the increased chance of terrorism and it helps a little with the problem of weight, as less hardware is needed. Additional important things to research next for this new idea are, the concept of electrical bicycles which resemble the suitcase in some ways, other applications of engine-assisting robots, user's preferences through a questionnaire and how to incorporate a smartphone application with Bluetooth connection in such a system.

Electrical Bicycles

An interesting application that uses similar technology is the electrical bicycle and it is useful to get more information about this technology. The basic idea is that the riders' input from the pedals influences the motor and via this the drive wheel of the bicycle. If the user is still able to have manual control over the bicycle, this system is a parallel hybrid system. In the case of a suitcase, the riders' input from pedals is changed to input from the smartphone application and the pull of the user to the suitcase. In bicycles, both brushed DC motors as well as brushless DC motors can be used and the main other components are just gears, hubs, rotors and rods. There are generally two types of electrical bikes; one in throttle mode in which you can set the amount of power the battery supplies with a throttle on the bicycle handles (like a scooter) and one in pedal-assisting mode, in which the battery helps you with pedaling. This last 'pedelec'-system comes in two varieties: one with a torque sensor pedal assist which measures the amount of power you deliver and emulate a certain percentage of that, and one with a cadence sensor pedal assist which supplies a constant amount of power which the user can select beforehand.

A fairly large difference in electrical bicycles and the intended suitcase is the battery used, since most electrical bicycles use either Lead acid or NiMH batteries. Both are mostly suitable for big devices which should deliver a lot of power and these are also banned by most airlines due to safety regulations. Luckily Li-ion batteries will probably be perfectly able to provide the energy needed for such systems.

Airport Regulations

There are several airport regulations when it comes to luggage weight, size and type of battery. Airlines are already putting a ban on several of the already existing "smart" bags, because they do not meet the requirements of the airlines. Which indicates the importance of these regulations when making a new smart luggage robot.

Handluggage

• 1 item of hand baggage, max. 55 x 35 x 25 cm (21,5 x 13,5 x 10 inch)
• 1 accessory, e.g. a handbag, briefcase or laptop, max. 40 x 30 x 15 cm (16 x 12 x 6 inch)
• Total weight max. 12 kg (26 lbs).

Check-in bagage

• 1 item of check-in baggage*: L + W + H max. 158 cm (62 inch)
• Total max. 23 kg (50 lbs).

Battery regulations

• Lithium Ion ≤100Wh
• Lithium Metal ≤ 2g

Research articles, patents and other relevant sources

A list of relevant sources (scientific papers and patents) follows underneath:

• Patent for self-following vehicle. This patents is actually pretty short and concise. It refers to other patents as well; it combines several systems into the concept of a target following devices. These systems are: a frame with a front and rear side, wheels, a driving system for wheels, a control system, a remote target unit and a stereoscopic detection system.
• Patent for a small DC motor. DC motors have of course existed for a long time, but this patent is specifically about a mini-sized DC motor. The important components of the DC motor are: a motor frame with a cylindrical portion with constant thickness, field magnets and a minimum sized air gap between the magnets and the frame to be able to rotate the armature assembly.
• Research paper on the design and implementation of a mapping robot with omni-wheels and a Raspberry-Pi. While the aim of this robot is different than in this case, the general architecture of this robot is still useful. Some requirements that are overlapping are: lightweight, movement at walking speed, enough space for sensors and electronics and relatively cheap materials. It was decided that four omni-wheels would be used for their mapping robot with one electric motor on top of each, as well as two double sided L298 motor drivers, an Arduino based controller, a Raspberry Pi as on-board computer and a Li-Pol 7.4 V battery as power source. The sensor that has been used is an ultrasonic HC-SR04 distance sensor. The reason why this article might be useful for this project is their conclusion that it is perfectly possible to create a robot with very general components which aim resembles the aim pursued in this project. The major difference is the weight it should handle and the autonomy of the robot, which can be tackled by using bigger motors, batteries and a stronger on-board computer.
• Research paper on the power and propulsion system for an autonomous robot. This paper is written about a robot participating in the RoboCup competition, a major robotics competition which has the goal of creating a robot soccer team which should win from the human team in 2050. The competition is attended by hundreds of teams each year and the robots get better each year. In this paper, the optimal energy storage system and propulsion system are researched. It is concluded that a Li-polymer battery of 25.9 V assisted by two Li-polymer batteries of 7.4 V each are the optimal power sources, since the power density is high. For the motor system, two cascaded cell modules for controlling motor speed and power flow control are used with DC-DC converters and a kicker circuit. The motor itself is a permanent magnet DC motor. An important remark made in the paper is that deep discharge needs to be avoided for safety and protection of the system. Thus, a basic voltage needs to be present so that the battery life is extended. Furthermore, for this project it is important that the batteries follow airport security guidelines.
• Research paper on a person-following autonomous trolley. This paper was written about the design of a robot that follows his user using both ultrasonic sensors and Bluetooth connection. The robot also reacts on human speech. The paper is mostly focused on the control style variants for robots. In many ways, this paper resembles this project, since it is about self-following robots used in mainly supermarkets, but use for disabled people, sports and military is also suggested in the paper. The approach of the researchers is different than the approach in this project however, since they consider 3D sensors and image processing not cost efficient. Ultrasonic sensors are used instead. While this can certainly work in supermarkets, it might be more difficult to make it work in busy areas as airports. Furthermore, an important remark is made, namely that GPS might be too unreliable at these distances. The robot made for this research paper mainly has components which are also present in similar robots found in other papers, namely two DC motors, a L289N motor controller, a Bluetooth module, an Arduino Mega on board computer, a 12V battery and regular mini wheels.
• Research paper on a novel object avoidance which is easy to tune and takes into consideration the field of view and the nonholonomic constraints of the robot. Moreover the method does not have a local minimum problem and results in safer trajectories because of its inherent properties in the definition of the algorithm. The algorithm is tested in simulations and after the observation of successful results, experimental tests are performed using static and dynamic obstacle scenarios.
• Patent for an object detection system. The method extracts a first feature vector from a first region of an image using a first subnetwork and determines a second region of the image by processing the first feature vector with a second subnetwork. The method also extracts a second feature vector from the second region of the image using the first subnetwork and detects the object using a third subnetwork on a basis of the first feature vector and the second feature. The three subnetworks form a neural network.
• Research paper on the linear time-invariant control system. The paper considers achievable delay margin of a real rational and strictly proper plant, with unstable complex poles, by a linear time-invariant (LTI) controller. The delay margin is defined as the largest time delay such that, for any delay less than this value, the closed-loop stability is maintained. Drawing upon a frequency domain method, particularly a bilinear transform technique, the paper provides an upper bound of the delay margin, which requires computing the maximum of a one-variable function.
• Research paper on object detection using convolutional neural networks for classification. This paper demonstrates that carefully designing deep networks for object classification is just as important as feature extraction. The paper also experiments with region-wise classifier networks that use shared, region-independent convolutional features.
• Research paper discussing the use of so-called pseudolites (pseudo-satellites) for close-range navigation. Using pseudolites, stationary devices that transmit GPS-like signales, and synchrolites, which rebroadcast GPS signals it receives from real GPS satellites, high-accuracy localisation methods become much faster and cost-effective, most notably Carrier-phase Differential GPS or CDGPS, which reduces the baseline error to just 1 centimeter. Localisation can even be done when fewer than 3-4 independent GPS signals are received, which is not the case when no other signals are received.
• Paper discussing how to achieve high accuracy using indoor pseudolites. Here, a constellation of GPS pseudolites is mounted on the ceiling of a large hangar, a car is fitted with antennae, and both run custom algorithms to determine the vehicle's location. The resulting error is less than 1 centimeter.
• Research paper on locating devices indoor using Bluetooth. This paper investigates the possibility of locating a device in a constrained environment using only Bluetooth. While this is possible, it requires many measurements to be taken of the Bluetooth signal strength from the `lost' device, while holding the receiving device (in this case, a smartphone) at specific angles and then move in the estimated direction of the lost device. This way, the location of the device can be reduced by a factor 4 with each measurement.
• Paper discussing Bluetooth beacons and a concept called stigmergy to locate devices indoor. Here, Bluetooth beacons are placed at known locations, and receiving devices can determine their own location using the received signal strengths from these beacons. In addition, it uses a stigmergic approach in which location estimates are treated like chemical markers (pheromones) in ant colonies, that diffuse over time and affect later location estimates.
• Research paper on localising devices indoors using Radio Frequency and Acoustic Ranging. In this paper, an algorithm is presented that uses Wi-Fi and acoustic signals to locate devices in an office environment. This requires that devices are equipped with Wi-Fi, a speaker, and a microphone. In turn, a localisation algorithm called EchoBeep can locate devices even in an environment with walls and other obstacles that obstruct and reflect signals.
• Research paper on distances between a robot and moving obstacles, such as humans based on the concept of depth space. To avoid obstacles the distance between the robot and obstacle is measured, based on the depth space and using an estimation of the obstacle velocity.

Force sensorless compliant control

Tech United, a student team at the Eindhoven University of Technology that develop robots that are used for playing robot football, among other things. These robots sense the environment using a camera that is pointed upwards at a parabolic mirror. This mirror reflects the incoming light from all around the robot, allowing for a 360-degree view with just one camera. Movement is done using three independently powered omni-wheels. These wheels allow the robot to move in any direction and rotate, all in one fluid motion. This combination of visual sensing and movement makes the robot a good basis for prototypes for many different kinds of robots. In the past, there have been cases in which a football-robot is equipped with different software to act as a proof-of-concept for other robotics systems. One such example is the work of Jerrel Unkel in : here, a robot is programmed to follow the movements of a person pushing the robot by hand. This way, the robot becomes easier to push than if no assistance was provided, e.g. if the robot's wheels could turn freely, but were not powered in any way. This concept of assistance is defined as follows:

Physically, the robot has certain properties that make it difficult to move: the most prominent examples of these properties are the robot's mass, which 'resists' acceleration, and its rolling resistance and other sources of friction. To make the device easier to move, the robot should act as if these properties are less of a problem than they really are: this means that the robot should 'feel' lighter and easier to move than it really is. Essentially, the robot has its own virtual mass and friction coefficients, which dictate how the robot should react to input forces. To do this, the device measures the disturbance caused by someone pushing it. The robot knows its own mass and friction coefficients, and can compute the movement caused by the user if no assistance is provided. Using the virtual mass and other properties, it calculates the movement that would be caused, if the device really had these virtual properties instead of the physical ones. Then, the motors are powered to the exact degree that the robot's movement corresponds to the movement according the virtual properties.

A simplified model would be as follows: suppose the robot weighs 10kg and ahs a virtual weight of 3kg. All friction forces are ignored here. If the user exerts a force of 6 Newtons on the device, the robot measures this and sees that the resulting acceleration should be 6N / 3kg = 2ms^-2. However, to accelerate the 10kg device at this rate, the needed force is 2ms^-2 * 10kg = 20N. As such, the robot itself should provide the additional 14N of force, so that the robot responds to the user's push as if it only weighed 3kg.

In , this concept is applied to the robot's mass, as well as multiple friction coefficients. Furthermore, the systems is extended to correct disturbances, e.g. differences between the real and desired position. These disturbances could be caused by differences between the theoretical and practical environment, for example the friction of the floor surface, or the mass if the robot. After all both of these values are not known intrinsically, so they have to be experimentally approximated. The robot does this by moving itself around in specific ways. For example, the static friction force is measured by slowly increasing the input force until the robot starts to move. For measuring the mass, the robot drives the motors in a sinusoidal fashion, measuring the movement resulting from this.

While this gives a very advanced and effective way of providing assistance, it would be completely overkill for our application. This is mostly due to the fact that this software is means for very heavy equipment: the robot it is tested on weighs 36kg on its own, while our suitcase will weigh no more than 10 or 12 kilograms at most. This has a number of important consequences:

• In , the assistance provided must be very accurate: if the motors do too little work, the user might not be able to move the device at all. If the required force is overestimated, this could lead to a system that is physically unstable.
• For heavy devices, the user cannot easily compensate for variations in the force required to move the device, meaning that the device itself has to be much more sensitive to variations in e.g. mass, friction, etc.
• These two factors combined mean that the device has to know exactly how much force is required to move itself in any particular way, at any particular time. By contrast, we know that the user can probably move the suitcase by him/herself without assistance, meaning that variations in the effort required by the user may be a little unpleasant, but certainly not disastrous. We, on the other hand, only really need to compensate the rolling friction, since the user will mostly be travelling at a constant speed.

Apart from the different requirements, our situation is completely different from a hardware perspective:

• The robot used in can move and rotate in all possible ways simultaneously, requiring measurements, calculations, and outputs in three different dimensions (movement in x and y directions, as well as rotation). Our suitcase can only move along one axis and in only one direction along that axis: towards the handle. Although the suitcase can also rotate, this is not done a lot and will not require any assistance.
• For us, it is important to minimise the total weight of the hardware, so we should use as few sensors, motors, etc. as possible.
• The user will almost always pull the suitcase by the handle, giving us one point at which we can easily measure the input forces. The devices mentioned in , however, could be of any kind and may just be pushed from any point, meaning that they have to rely on accelerometers and disturbance sensors.

In general, our suitcase is a context which places more emphasis on simplicity and low cost, instead of a general solution with strict requirements on accuracy. This is possible due to the relatively low weight of the suitcases, and the fact that they are always pulled from the same point, allowing us to make more direct measurements.

Questionnaire Results Analysis

A questionnaire was created with a total of 10 questions and spread online through social media in order to have the opinion of people worldwide. The number of participants is 100 and from various countries, the majority being from The Netherlands and Tunisia. The following questions were asked:

1. How often do you fly per year?

2. How often do you fly with only hand luggage?

3. Do you have trouble moving your luggage, for example at large distances?

4. In case of a yes, where do you have trouble with when moving your luggage?

5. In case of a yes, do others help you with moving your luggage?

6. Do you fear losing your luggage?

7. How important is the weight of an empty luggage for you when buying the suitcase?

8. Is it difficult for you to adhere to the maximum weight for your suitcase?

9. What functionality would you consider useful on a ‘smart’ suitcase?

10. Would you buy a suitcase with these functions?

Analysis of the results

Each question had multiple answers, most with only one correct answer except question 9 which had multiple possibilities along with a field Other in order to have more insight on people's opinion.

The most interesting questions to analyze are questions 4, 9 and 10. According to the data, it seems that when people do indeed have trouble with moving the luggage in long distances, they would like to have their hands free. The following image shows the results:

Moreover, question 9 also contains extremely interesting information: it seems that GPS location, smartphone connection and self-following are the top answers. Furthermore, in the Other field, 13 people added suggestions in which 6 are about displaying the weight of the luggage in the mobile application. The following image shows the results:

Last but not least, question 10 shows that by adding these features, 78% of the users are willing to buy the suitcase. The following image shows the results:

Conclusion

It goes without saying that the results are promising, showing what people from different cultures (mainly Europe/North Africa) think of such a concept. According to the data, it seems that a hand luggage equipped with GPS tracking and a smartphone application, which also makes it easier for the users to move it around would lead to the users buying it. The data from the questionnaire will be taken into account in deciding which features to add but also in the design of the suitcase itself.

Hardware & Design

One of the challenges of this project is to measure the amount of work done by the user in pulling the suitcase and converting this into motor output voltage. Two relatively easy systems exist for that, with one system measuring pulling force in the handle of the suitcase and another system in the wheel axis measuring the torque of the wheels when pulling the suitcase. The system is the handle is easier to implement while the system in the wheel axis is more accurate. The handle system is less accurate since it has to convert the pulling force to forward speed when the user wants the motor to produce a constant speed. However, in order to do this, the angle between the handle and the ground needs to be known, which varies per moment and per person. Also, when the user wants a certain relative amount of work to be done by the motor on the other hand, the system needs to convert the pulling force to a certain amount of motor power. This can be rather inaccurate, so another system in the wheels for measuring torque. This is a lot more accurate since it measures the actual forward speed of the suitcase directly but it involves also some additional instruments. However, since the handle system only really involves a basic electrical circuit (it can function as a weight measuring system as well, more about that later on) it is no problem to implement that as well.

Torque sensors

The most reliable way to let the motor produce a certain relative amount of power is to measure the torque in the wheels when the motor is still off. With the smartphone application, the user can select the amount of work wanted to have relieved and the torque sensors measure the torque on the wheels. The sensors send a signal to the computer system in the suitcase and depending on this the motor generates an amount of torque itself. Then, the sensors again measure the torque and compare it to the necessary amount of torque, so it checks whether the relative amount of work done by the motor is the right amount set by the user.

When talking about torque it is important to understand the difference between dynamic and static torque. Dynamic torque is torque which changes over time, so there is an acceleration present, while static torque is constant over time. In this case it is easier to talk about static torque; the sensors measure the torque on specific times and adjust the motor speed to that. Most of the times the speed of the user plus suitcase is constant anyway, so while it would be more realistic to treat the torque in this case as dynamic, it doesn’t add a lot to the system and dynamic torque sensors are a lot more complicated. Thus, a static torque sensor would be most useful.

Schematic image of a torque sensor

There is another important distinction to be made in the field of torque sensors and that is the difference between reaction torque sensors and rotary torque sensors. Reaction torque sensors are not quite useful to this project since the rotation here is limited to 360°. Rotary torque sensors are mounted underneath the device under test and have no rotation limitation. The actual measurement of the torque is done by either strain gages in a Wheatstone bridge circuit, optical speed sensors or a combination of both. Sensor input is recorded in Volt and transmitted to the onboard computer.

However, another useful function is to let the motor produce a constant, absolute amount of power. A sensor is needed for that that measures whether the user is pulling the suitcase, since the suitcase should not let the motor run without user's assistance. In electric bicycles there are cadence sensors which can measure whether the user is cycling on the pedals or not. In the suitcase it would be sufficient to just use a combination of the torque and the weight sensors; if both are measuring movement, this means the suitcase is in use and that the motor should produce a constant amount of power if that mode is selected. If only the weight sensor measures movement, it is probably just lifted from the ground and if only the torque sensor measures movement, this also means the suitcase isn't actually moving. Furthermore, if the suitcase is both lifted up and the wheels are turning, the motor would produce power, though this isn't really a problem and such a position would not be maintained over an extended amount of time.

It has to be noted however that such a torque sensor works better on electric bicycles, since there the amount of force the user delivers in order to produce movement can easily be tracked by using a torque sensor in the pedals of the bicycle. No matter how much power the motor delivers, the amount of force which is delivered by the user can always be measured by measuring the torque in the pedals. In the case of an electric suitcase, this is not the case. There are no pedals present and a torque sensor is not ideal for measuring the amount of force delivered by the user. In this case, there are two axes; one which is connected to the two wheels of the suitcase and one which is connected to the motor of the suitcase. When the user pulls the suitcase, the suitcase will move and the wheel axis will turn. This axis movement needs to be measured in order for the motor to switch on and deliver work as well. When the motor turns on, it will rotate the motor axis and via similar methods as in an electric bicycle it will transfer the motion of its axis to the wheel axis and the wheel axis will turn faster with the user still delivering the same amount of force. Now still the amount of force the user delivers can be measured via the rotation of the wheel axis, but now the additional motion caused by the motor needs to be taken into account; this has to be subtracted from the total amount of rotational motion of the wheel axis in order to measure the amount of work done by the user and this is needed to check whether the motor indeed delivers the desired amount of work. However, a sensor is needed that measures this rotational motion, and this sensor is called the Hall effect sensor.

Hall effect sensor

A hall effect sensor is a transducer which is able to measure changes in the magnetic field present by changing its voltage. The main application for which a hall effect sensor will be used in this case is for measuring the speed of several axes. Hall effect sensors are often incorporated in (brushless) DC motors as a kind of feedback sensor. While the input voltage of motors should already determine the speed of the axis of the motor, there might be issues and it can be useful to have a hall effect sensor inside the motor. However, it is in this case very important to have a hall effect sensor inside the axis which is connected to the wheels, as this is the only way to measure what the speed of the velocity would when only the user’s force is considered reliably, when the motor is also on.

Schematic image of a hall effect sensor

Hall effect sensors are triggered by an external magnetic field which is variable in time and this is easily accomplished by letting the axes have a gear shape instead of a perfectly round shape. This way, the distance between the magnet in the hall effect sensor and the axis varies with time when the axis is turning and with the frequency with which it turns the rotational speed of the axis can be determined.

Weight sensors

It is quite convenient and not very complicated to implement a weight-measuring device into the smart suitcase. A simple way of measuring weight is by using a spring like in a normal kitchen scale. This method uses Hooke’s law to give a good approximation of the force on the weight scale and thus of the weight of the object on the weight scale. It doesn’t have to be a digital weight scale in this way, as the output can be given on an analogous scale and no electricity is needed for that. The downside is that there are some inaccuracies; the calibration may not have been done properly, Hooke’s law is only a linear approximation, the spring used may wear over time and it is not easy to read the output accurately. Furthermore, when using electrical components like in the suitcase, it is easier to just switch to the digital variant, which does not use spring but rather an electrical circuit named the ‘Wheatstone bridge’ which is known for its extreme accuracy and is used in a variety of applications. . A Wheatstone bridge consists of a number of resistors, usually four, of which two are known are known and constant, one is adjustable, and one is unknown. This unknown resistance varies when force is applied in the right way to the circuit. By adjusting the adjustable resistance and using a galvanometer, it is possible to find the value of resistance in the adjustable resistor in order to let the galvanometer show a zero reading, meaning equilibrium. Using calibration this value of resistance can be used to calculate the force applied and thus the weight. This calculation as well as some needed noise-filtering is done by a small computer that fits into the weight sensor. Digital weight sensors like kitchen weight scales use such methods and it can easily fit into the handle of a suitcase.

Batteries

Before looking into the subject of batteries more deeply, first the findings on airline regulations need to be recapped, since this has the highest priority for the actual product. Batteries that are forbidden here are completely out of the question. Firstly, any batteries in any device need to be removable from the device and there are very strict battery regulations for checked luggage. The types of batteries that were allowed were:

• Dry cell alkaline batteries (AA, AAA, C, D, 9-Volt, button cell batteries)
• Dry cell rechargeable batteries (NiMH, NiCad)
• Lithium ion batteries (Up to 100 Wh, 101-160 Wh with approval)
• Lithium metal batteries (Up to 2 grams of lithium per battery)
• Nonspillable wet batteries (Up to 12 V, 100 Wh per battery)

In most of these cases the amount of batteries which can be transported is limited, sometimes to two, sometimes to more than twenty. The most important battery type in the list is the type of Lithium-ion batteries, since they are used most in medium-sized robots. Most laptop-batteries are allowed for example and they can roughly deliver the same amount of power as is needed in a smart suitcase.

Some types of batteries that are not allowed or for which sharp restrictions exist are:

• Lead acid batteries
• Car batteries
• Industrial batteries like absolyte, NiFe-batteries or steel case batteries
• Nuclear batteries

Most of these batteries are either too large or too heavy for the intended purpose anyway. Most of the times, such batteries are shipped or transported overland rather than via airplanes.

Then there is the option between one or multiple batteries. The main reason for the difference is that difference parts of a robot need different voltages: sensors at approximately 5V, motors at 9V-12V but servos at 3V-6V and other electronics at 9V-12V. When using a simple design, it might be better to just use different batteries at different voltages for the different components. However, all these batteries need to be removable and need to be removed at border control, as well as all the batteries need to be recharged separately. Furthermore, using microcontrollers the voltage can be adjusted. Thus, in this application it is more useful to use one main battery.

A popular choice is the lithium-polymer battery or LiPo for short. It has a rather high specific energy and higher than comparative batteries. For the same amount of energy, the battery is thus lighter. Since weight is an important feature in the suitcase, this is very useful. Furthermore, they are relatively unconstrained and somewhat flexible. Some laptops use LiPo batteries as well, as well as some electric vehicles. However, LiPo suffers from the same risks as other Lithium-ion batteries, namely explosion and fire hazard. These can happen when the battery is overcharged, is either very hot or cold, when there is a short circuit or when it is penetrated by a sharp object. The electrolyte then leaks, causing explosions or fire. Several instances of this happening have been recorded. This risk should of course be minimized and a number of precautions are made on the more advanced batteries, such as over charge protection, EMF protection, over discharge protection, thermal effects protection, safe reset mechanisms and general over current or over voltage protection. These are called ‘protected batteries’.

Wheels

The initial idea for a self-following suitcase was to put four omni-wheels under the suitcase which were motorized individually. Research on state-of-the-art showed that some airlines had prohibited self-balancing devices like segways for transportation on airplanes and that potentially more had this intention. It was thus impossible to make an autonomous device that had just two wheels, since it should be able to balance itself when in autonomous mode. However, when the switch to an engine-assisting suitcase was made, some assumptions changed. First of all, the user always has a hand on the handle of the suitcase and thus always has a physical connection with the suitcase. A consequence of this was that the user would automatically provide the balance for the suitcase, as the user is prone to do with normal suitcases anyway. The option of two-wheeled suitcases was then open again. Additionally, torque sensors were needed for the wheel system. It is then much easier to actually implement the wheels into the suitcase, rather than have them completely under the suitcase. Furthermore, it is quite complicated to make an engine-assisting robot with wheels that can turn 360 degrees and many errors can be made with this. Lastly, one wins a significant amount of baggage space when the wheels are incorporated into the suitcase itself. Thus, two wheels should be used in this suitcase which are implemented in the suitcase itself. These wheels can be regular robot wheels, though there is the requirement that they can be powered by a motor. Also there is the requirement that they also should be able to work when the motor is switched off, but this is more of a requirement for the motor. There is no need for the wheels to be operated separately, as the suitcase will not be remotely operable or be able to turn on its own. There is also no need for two additional wheels at the other side of the suitcase, as the handle of the suitcase is just on one side and the suitcase will be pulled from one direction only.

Motor

Several different motors exist for robotic applications. The most important are:

• AC Motor
• Brushed DC Motor
• Brushless DC Motor
• Geared DC Motor
• Servo Motor

From these, the AC motor is definitely the least convenient. Most electronics for robotic application works on DC (direct current) and so do most batteries which can be used for robotics. Furthermore, most AC motors are rather big and way too powerful for the intended purposes. Secondly servo motors are not particularly useful either. Most of them are limited to 180 degrees of motions and are not strong enough to move a load of 10 kg for an extended period of time. This leaves DC motors, but there are some varieties in this category.

After some reconsideration, it seems most advisable to use brushless DC motors with additional gears for lower power levels. The main difference with brushed DC motors is that these motors produce an audible whine sound from the commutator brushes, that brushed DC motors also have a lot of electrical noise which can damage electrical components and that brushed DC motors are not very efficient. Brushless DC motors don’t have these problems. However, brushless DC motors need a separate motor controller and are slightly more complicated to operate. Additionally, a number of gears is used to easily switch between lower and higher speeds. The plain brushless DC motor doesn’t operate too well on low speeds. A simple gear construction can be used to switch between speed and torque. .

The basic working principles of a brushless DC motor are fairly simple. There is only one moving part and that is the rotor. It is rotated by energizing the windings inside or outside of the rotor in a specific sequence. The idea is to constantly attract the north and south poles of the rotor to the direction the roto should rotate. Therefore, the north and south poles of the windings need to switch constantly. In this manner, the mechanics are similar to that of a particle accelerator. It is then also important to incorporate acceleration in the system. The idea is that the rotor spins at a certain constant speed if the magnetic poles of the external windings switch periodically. If this period is shortened, the rotor will spin at a higher speed and vice versa. The process of shortening the period causes acceleration of the rotor. Control is also needed in this case, more than in comparative motors. Most brushless DC motors use a so called ‘Hall Effect Sensor’ or an optical sensor to detect the position of the poles of the rotor.

Actually, most brushless DC motors are a combination of several AC motors. The system produces an AC power output of a certain phase from an DC power input from the battery and this rungs the brushless motors by sending this sequence of AC signals originating from the electronics speed control’s circuitry in the onboard computer. All of this combined results in no sparks, long lifetime of components, few noise, few electrical noise, high efficiency and high maximum speed.

List of components

• Two LiPo batteries
• Two motor-powered wheels
• Two brushless DC motors with an additional gear box
• Robot motor controller
• Arduino or RaspberryPi as microcontroller
• Wires
• Static torque sensors
• Weight sensor with Wheatstone Bridge
• Bluetooth sensor and transmitter
• GPS sensor and transmitter
• Battery life sensor (Voltage sensor)
• Temperature sensor to prevent overheating
• Strong polymer case
• Extensible handle

Assistance method

For assisting the user in moving the suitcase, we imagine several different options:

1. Firstly, the motors could push the suitcase in the direction measured by the sensors with a constant amount of force. This amount can then be adjusted, e.g. using the app. The main advantage of this method is that it does not require that the suitcase measures the magnitude of the force applied by the user: only the direction has to be known, which helps simplify the sensors as well as the software. It may, however, lead to situations in which the suitcase moves with more force than the user applies, causing it to 'shoot ahead' of the user. When the suitcase is not being moved, it is important that the sensors don't detect 'noise' caused by small disturbances or unevenness of the floor, as this could result in the suitcase moving randomly by itself. One way to prevent this, is to deactivate the motors when the user is not touching the handle. Since there is already a sensor here, this could easily be implemented. However, it would also mean that the user can't push or pull the suitcase without using the handle.

2. A second option is to define the force exerted by the motors to be a certain (adjustable) percentage of the total required force. For example, the user could set the suitcase to always do 70% of the work, with the user only having to do the remaining 30%. This method may feel more 'natural' to the user; for example, if a suitcase with a 10kg net weight assists for 70%, it may feel like it only weighs 3 kilograms to the user. However, it will also require more complicated sensing and software. Specifically, it may prove difficult to measure the torque from the user while the motors are also turning the wheels. Of course, it will also require that the magnitude of the force is known, not just the direction. Furthermore, this method is limited by the maximum torque of the motors, meaning that if the user applies too much force on the suitcase, the torque required from the motors has to be limited in some way, so as to prevent damaging the electronics. In general, this method may prove to be more demanding on the suitcase, since there is no theoretical limit to the amount of force that will be needed.

These two options all have advantages and disadvantages, and requires us to compromise between complexity and user experience. However, we note two things:

• Even when the magnitude of the sensors is required, we expect a modern microprocessor like a Raspberry-Pi or an Arduino to be more than capable of processing these inputs quickly, meaning that the software complexity is probably not a concern.
• Secondly, measuring the magnitude of a force should not be very complicated either, since force sensors and torque sensors are fairly common and widely available.

Together, this means that for the first option, the only significant concern is the maximum load on the engines. Therefore, we propose a third method for assisting:

3. A third option would be to use a combination the above approaches: as the user starts applying force, the suitcase compensates a set percentage of it. As the user's force increases, however, the suitcase's contribution approaches a certain maximum that it does not exceed. This way, the suitcase's behavior is still natural and predictable in most cases, while ensuring that the system is not 'overloaded' in any case. However, the effective behavior is then that the suitcase appears to 'resist' going faster: it will compensate a smaller part of the required force, meaning it will feel heavier. The user will interpret this as meaning they should be going more slowly, which means it's very important that this does not start happening at low speeds.

This last method of assisting looks the most promising to us right now. However, all three options should be experimented with when a real prototype is being developed.

Working principles of prototype

Force diagram for the handle of the suitcase

Force diagram for the wheels of the suitcase

Force diagram for the suitcase

In the figures in this section some force diagrams can be seen as well as force sensors incorporated in the suitcase which illustrate the basic principles of the suitcase. CM is the center of mass of the suitcase, F(U) is the force the user delivers for moving the suitcase, F(G) is the force of gravity pointing straight downwards, F(M) is the magnetic force present in the motor which rotates the wheels, F(N) is the normal force applied from the floor to the suitcase, F(f) is the generic friction force in the opposite direction of movement, F(f,r) is the roll friction force and F(f, ds) is the dynamical sliding friction force. The angle α represents the angle between the direction of force from the user and the direction straight up vertically, the angle β represents the angle between the long side of the suitcase and the direction straight up vertically. In the pictures a few wiggly grey diagrams can be seen. These are the Wheatstone bridges which are extended slightly when force is applied to it. This extension is proportional to the voltage and the voltage is delivered to the on-board computer as data. One of these Wheatstone bridge systems is present in the handle of the suitcase and one is placed parallel to the handle while one is placed perpendicular to the handle. These Wheatstone bridge circuits thus measure the forces in respectively the direction parallel to the handle and perpendicularly to the handle. There is also a Wheatstone bridge system present above the wheel axis. These circuits measure the forces which are on the axis of the wheels, also component-wise (the same components as previously). These forces are also converted to data in voltages and combined with the known value of the force of gravity, the complete force diagram can be computed by the on-board computer. This is important, since this way the on-board computer is able to determine the force the user applies in order to move the suitcase forward.

An additional advantage of the system is that it can take possible slopes into account. This doesn’t happen with a slope detection system, but it rather happens automatically. When you are moving with the suitcase on a slope, the force of gravity either works with you or works against you, since one of the components of the force of gravity is in the direction of motion or against the direction of motion. When you are walking down a slope with your suitcase and you keep your speed the same as when you were walking on a flat surface, the suitcase can detect with the Wheatstone bridge circuits that the total force externally applied in the direction of motion is the same, but now it is the sum of the force of the user and some part of the force of gravity. When you are walking up a slope and you keep your speed constant with respect to the situation without slope, the system will just detect a lower amount of force in the direction of motion (namely, the force of the user minus a certain percentage of the force of gravity. Alas, for the system this change is just the same as going slower on a flat surface and while the angle between and the suitcase and the floor will change on a slope with respect to the situation without slope, this can also be due to changes of position on a flat surface. Additionally, the direction of force from the user with respect to the horizontal and vertical plane (i.e. the horizon) will most likely not change, so no change can be detected here. However, there will be a change in gravity along the component perpendicular and parallel to the slope. A sudden change in this component can be detected and it can be made possible that the system automatically increases the relative amount of power the motor delivers, such that walking on a slope with a suitcase feels the same for the user as walking on a flat surface with a suitcase.

Lastly, it is very important to know how exactly the collaboration between the motor and the user goes. As was said earlier, it resembles the working principles of an electric bicycle, but there are some differences. The use of the torque sensor was already mentioned, but it is also the case that with an electric bicycle the user is rotating an axis, the motor is rotating an axis and these rotations are combined in one final axis connected to the back wheel (in most cases). With the suitcase this is slightly different; the user is pulling the suitcase, thus rotating the axis of the wheels directly. The motor axis is separate of this axis. In a way, the situation of the suitcase is more similar to that of an electric skateboard. Namely, here also the user uses his energy to rotate the same axes as the wheel axes and the motor axis is separate and assists the rotation of the wheel axes. However, it is important to note that in all cases, the axis of the motor is rotating at a different speed then the axis or axes of the wheels. Luckily, this can be made possible easily with the use of several disengaged clutches and a number of gears, such that the motor is indeed merely assisting the user and not completely taking over the work of the user. Now only the issue of the control system rises, since the motor speed should be dependent on both the amount of work the user delivers and the amount of work the user wants to be relieved off. There should be a feedback system that constantly checks whether the motor speed is correct. For this, an ingenious control system with a feedback loop and a transfer function is needed and the computational work is done by the on-board computer. We will not discuss the details of this here.

Pseudocode for calculations

Below, pseudocode is given for the algorithm that determines the force that should be applied to the wheels, as well as a small method that can be used to weigh the suitcase. There are a couple of things to note about this code:

• The software eventually sets the required force on the wheels. It may not seem obvious that this is possible, since electric motors usually have a variable speed, not torque. However, this is not a problem here. Since the radius of the wheels is known, the required force can be converted to a torque. The suitcase uses a brushless motor, which has a known torque constant expressed in Newton-meters per Ampère. The current can be controlled using the microcontroller that is part of the motor.
• When weighing the suitcase, more accurate measurements can be achieved if the force on the handle is measured multiple times. This is not represented in the code, since it would require interaction with the smartphone app (e.g. an instruction to lift the suitcase up by the handle). So instead, only the basic way of making a single measurement is given here.

Design Decisions for Hardware and Design

There is a certain amount of design decisions that need to be made and the user’s needs are crucial in this process and need to be considered. For this, it is good to recap what our target user group was again. Among these groups like disabled people, elderly people and generally people who have trouble with transporting goods over large distances exist, as well as businessmen or people who travel regularly. One of the decisions which have been made is that there should be two methods of assisting; one with a constant amount of assistance from the motor in the robotic suitcase and one with a relative amount of assistance. The first is useful for people who are either in a hurry, since the robot always provides the same amount of force which results in the same amount of base speed and when the user decreases his amount of work done the speed of the suitcase will not decrease much compared to when the relative amount of assistance is used. This mode may be more difficult to adapt for some people, but the battery life is longer and it will probably run smoother. These modes are also present on electric bikes and users are satisfied with these modes, since for both modes there are people who love it and hate it though there is not a need for a new mode. However, this might be different for the suitcase ide and it would be very useful and informative to do user tests on this subject.

Another important technical design decision was to let the suitcase have two wheels and not four wheels. Both options are used widely in the baggage industry and both have their advantages and disadvantages: the four wheeled suitcases are more handy when there are several bags, it can function as a cane, it is easier to push while in a queue and it can function as a two wheeled suitcase by just tilting it, however the wheels are often not built in but rather just put under the luggage, which makes them more vulnerable and it creates a fair amount of empty space which counts for baggage limits. If this is not the case, it still takes out twice as much space since there are twice as much wheels. A two wheeled suitcase however is more useful for larger speeds of the user and since the wheels are built-in the don’t go to unwanted directions, which can happen with wheels that can turn in every direction. This is to say, maneuvering is easier with a two-wheeled suitcase. Additionally, the design is usually a bit sturdier than a four-wheeled suitcase and it is easier to pull a two-wheeled suitcase over uneven terrains. However, you do have to use a certain amount of force to compensate gravity which is not the case with a four wheeled suitcase.

These are all general considerations and in our case there are certainly some priorities. Firstly, it is the case that with our product a fair amount of space in the suitcase is taken up by electronics, at least certainly more than with regular suitcases or even regular smart suitcases. In that case, a two wheeled suitcase is better for this design, referring to R27 of the requirements list. Furthermore, there are quite some electronics needed for the wheels and the wheel system, an important part of that being the motors. This is not easily hided under a suitcase and with a four-wheeled design there is a lot more electronics needed, which takes up weight as well. Weight is one of the other priorities and this should really be minimized, referring to R27 of the requirements list again. Also, creating a force model is easier when there is just one way the user can pull the suitcase, which is the case in a two-wheeled suitcase. A four-wheeled suitcase is more free in this manner which makes things complicated. Assistance will also work a lot better when the wheels are fixed in the suitcase and have only one direction to turn, referring to R8 of the requirements list. However, it is certainly the case that for disabled people it might be more comfortable to also be able to push the suitcase or lean on the suitcase while walking. Ultimately, it seems to be the case that a two-wheeled device is more advantageous, certainly in the beginning phase of such a product. If technologies advance, it might also be possible to create a four-wheeled suitcase for a reasonable price.

Software & Technology

It goes without saying that our system needs multiple technologies in order to be useful for the users and for society. Such technologies include GPS tracking in order for the user to be able to locate his/her suitcase when stolen or lost and bluetooth connection in order to access information about the luggage such as battery life. Moreover, security is of great importance where a fingerprint lock can be added as a two-factor authentication in order to ensure that no one can stole the content of the suitcase.

GPS Tracking

In order for the user to track the position of the luggage at all times and locate it if the system gets stolen or lost, a Global Positioning System (GPS) tracker is used. This technology uses a receiver that communicates with satellites to determine global position that is accurate to up to 5 meters, which is enough for a user to see the suitcase. A GPS tracking system uses the Global Navigation Satellite System (GNSS) network. This network incorporates a range of satellites that use microwave signals that are transmitted to GPS devices to give information on location, vehicle speed, time and direction. So, a GPS tracking system can potentially give both real-time (active tracking system) and historic navigation (passive tracking system) data on any kind of journey. The control of the Positioning System consists of different tracking stations that are located across the globe. These monitoring stations help in tracking signals from the GPS satellites that are continuously orbiting the earth. Space vehicles transmit microwave carrier signals. The users of Global Positioning Systems have GPS receivers that convert these satellite signals so that one can estimate the actual position, velocity and time. For our purpose, we are more interested in real-time data and thus the active system as the user wishes to locate the suitcase in the current time.

This data can then be sent to the mobile application on the user's phone.

Bluetooth Connection

During normal use, the user must be able to interact with the suitcase. An example of such an interaction is the user monitoring the amount of charge that remains in the batteries. Many of the features described under the section 'security' will also require that the suitcase and its user can communicate with each other. For this, we will use a Bluetooth connection between the suitcase and the user's smartphone. The connection will be initialized using a smartphone app, that will be described in further detail under the section 'Mobile Application'.

Our choice to use Bluetooth was largely based on , which compares four popular wireless data transfer protocols: Bluetooth, ultra-wideband (UWB), ZigBee, and Wi-Fi. These protocols are compared by transmission time and throughput, data coding efficiency, protocol complexity, and power consumption (both overall and relative to throughput). Wi-Fi and UWB have very high throughputs and equally high power usage, meaning that in our scenario, where the required throughput will not be very high, these protocols are not very suitable. Comparing ZigBee and Bluetooth, the paper notes that the ZigBee protocol is much less complicated, being defined by the IEEE using just 48 primitives compared to Bluetooth's 188 primitives. This would make ZigBee more suitable for devices with little processing power such as our suitcase. However, the power usages of both systems are similar, and Bluetooth consumes significantly less power per Mb of data sent or received. Since both protocols have low throughput (less than 1 Mb/s), this means that Bluetooth may consume less power than ZigBee in our use case. However, as mentioned before, the differences between both protocols in the areas that are relevant to us, are quite small. We elected to use Bluetooth for our suitcase, since this protocol is widely available on smartphones and is already familiar to many users.

A more detailed analysis of Bluetooth's power consumption is given in . This paper gives a detailed method to measure the power consumption of Bluetooth Low Energy (BLE) devices. It also uses this method on an example device, namely a keyfob with some buttons, representing a generic device with little processing power that will occasionally transmit and receive little bits of data. It makes the following measurements: a connection event, which consists of a data packet being received, and a response being transmitted, takes between 2.6 and 2.7 milliseconds. This time includes the time it takes for the Bluetooth chip to wake from its sleeping state, receive the message, transmit a response, process the message, and go back to sleep. The average power consumption for one such event is measured to be between 8.25 and 8.53 mA, with 0.001 mA being used when the chip is in sleep mode. The average power consumption with 1 packet being received per second is then equal to 0.0230 to 0.0247 mA, meaning that the device could receive 1 message per second constantly for approximately 400 days when powered by a standard coin-cell battery with a 230mAh capacity. Note that in these tests, the packets sent are empty (e.g. have a full header but empty payload). In practice, meaningful messages with an actual payload would mean that the time to receive, transmit, and process will be higher, and power consumption will therefore also be higher. In our case, the messages should still be small (single numbers representing e.g. battery percentage), so we expect this paper's results to be a good indication of our real-world power usage. Furthermore, the measured power consumption of the Bluetooth chip will then be insignificant compared to that of the electric motors, meaning that 'stand-by' power usage will not be a concern for us.

Calendar

Another feature that can be added is for the suitcase to remind the user to charge the system one day before traveling. In this sense, the suitcase should synchronize with the calendar of the user. In case the user plans to travel the next day, then the mobile application can show a message or notification in the user's phone suggesting that he/she should charge the suitcase. In this way, the user avoids having an empty battery in the airport and can use all the features.

Security Systems

Two-factor authentication: In order to ensure more safety, a two factor authentication can be added to the system. Also known as 2FA, this method is an extra layer of security that is known as "multi factor authentication" that requires not only a password and username but also something that only, and only, that user has on them, i.e. a piece of information only they should know or have immediately to hand - such as a physical token. It is already commonly used in mobile applications such as WhatsApp. When implementing this extra security layer, one needs to take into account the policies in the United States, namely the Transportation Security Administration (TSA) lock in which the US board control should have the possibility of opening the suitcase. In this sense, the system could have a physical lock accompanied with a lock on the mobile application (either pin code or fingerprint). However, in the scenario where the user loses the phone or the phone is turned off, then the baggage cannot be opened until the phone is found or turned on again, which is not desired. Another possibility is to have at all times a fingerprint system on the baggage itself, however this would not allow the United States borders to open the baggage if needed. Hence, we opted for a two factor authentication that would use a fingerprint system on the suitcase along with a TSA compliant physical lock, with the fingerprint system that can be disabled from the mobile application. In this way, the user can disable the fingerprint system when entering the United States customs.

Position Tracking: Using the GPS tracking system discussed earlier, one can always track the position of the system using the mobile application. If for instance a suitcase is stolen in an airport, then the owner can use the mobile application to find the suitcase and ensure that the content is not stolen.

App Development

Need for an App

A mobile application is necessary to accompany the smart suitcase. Even though it does not directly influence the engine assisting control system of the suitcase, features such as battery percentage and GPS tracking help users plan their trip and enhance security. These features can either be displayed on an on-board screen on the suitcase itself or a smartphone application. By using a mobile application, it makes it much easier and simpler for the user to access the informations at all times. Today, 2.53 billion people own a smartphone , meaning that such an application is widely accessible worldwide. The data gathered by the questionnaire shows that that users would like to use a smartphone application connected to the suitcase (52% of the results on question 9 of the questionnaire) and use features such as GPS and weight display that can be easily accessible on a smartphone. An alternative would be to use an on-board screen on the suitcase itself. However, this solution is not practical since one needs to consider situations where the suitcase is stolen. If such a scenario were to happen, how would the user locate the suitcase? The GPS functionality would then be useless. Moreover, since suitcases are exposed to possible turbulences in the airplanes, the screens can easily be broken, thus making all the features inaccessible for the user.

The mobile application contains 6 features listed below:

Track the position of the system: Using GPS, the application can locate the position of the suitcase. GPS can be theoretically accurate up to 5 meters, which is more than enough for the user to see where the suitcase is.

Display weight: Using the weight sensor on the handle of the suitcase, the application can display the weight in kilograms to the user.

Display battery life: By communicating with the on-board computer, the suitcase can display the battery life in percentage to the user. The battery life is calculated with the amount of hours left and how many kilometers the user can still walk with the suitcase.

Select the assistance method desired: The user can select whether to use Absolute speed or Relative speed, as detailed in section Assistance method.

Setting the security system: The two-factor authentication using a finger print can be enabled or disabled by the user.

Notify the user one day before the trip: The mobile application is synchronized with the calendar of the user and notifies him one day before the start of the trip as a reminder to charge the battery.

User Interface

The home-screen is the first application page that appears when the user starts the application. It consists of 4 blocks that redirect to the other activities in the application. The first block is the battery life that shows the percentage of battery of the suitcase [R17]. The second is a weight block that shows the last weight measurement [R16]. The third block is the settings block that redirects to the settings activity and finally there is the find case block that redirects to the find my suitcase activity.

The find case page shows a map with the location of the suitcase. There are two buttons that enable the user to zoom in and out on the map. In the middle there is a button that targets the suitcase on the map and centres that location on screen. [R24].

The battery page updates the battery life and shows how long the user is able to walk with the smart suitcase on and how much time is left before the suitcase needs to be charged [R17] and [R18].

The setting page has a button to turn the suitcase on and off [R21]. It shows an overview of which suitcase is connected and also the suitcases with which the app has connected in the past [R20]. It has two option to choose the movement protocol, namely absolute speed or relative speed [R22]. Finally there is a button to turn the security lock on the suitcase on or off [R21].

The weight page shows the last measured weight of the suitcase, which was also visible from the main menu, to quickly see if the suitcase adheres to the weight limit [16]. It is also possible to update the weight whenever you have taken something from the suitcase or placed something inside, by pressing the buttons displaying the scale for a new measurement.

System Specifications

Battery

As was said earlier, there are certain kinds of batteries which cannot be used in a suitcase used for air flight and the total energy stored in the battery should not be higher than 100 Wh. For batteries between 100 Wh and 160 Wh, most batteries need approval. This is of course not desired. Considering the amount of energy stored in the batteries of the smart suitcases of the companies mentioned in the theory (the ones making the autonomous suitcases), which is around 100 Wh, and the average battery life of these products, which is around 4 hours, it seems likely that such a suitcase as ours would function with batteries with a total amount of energy stored of 100 Wh, since the autonomous suitcases have more components, more functions and generally consume more energy (for for example the object detection and avoidance system). Thus the best idea is to meet the limit of 100 Wh to avoid hassle at airports for users. However, the prices of batteries are decreasing rapidly. Furthermore, for users it is favorable if the battery lasts long and they will only use it for a few hours straight afterwards the battery can be recharged for a long time (which is of course needed for large batteries). Thus, it seems logical to meet the limit of 100 Wh, but just barely, and let the battery have a total amount of energy stored of about 95 Wh.

Additionally, there are a few things to be considered when choosing for a certain type of battery. The main considerations are:

• Weight
• Safety
• Environmental friendliness
• Speed of power supply
• Power density (i.e. size for 95 Wh)

Below, a table is put in order to compare the main battery types on these considerations:

Table 1:

Battery Type	Weight of 95 Wh energy storage (kg)	Size for 95 Wh energy storage (L)	Voltage per cell (V/cell)	Price for 95 Wh energy storage (€)	Environmental Friendliness
LiPo battery	0.36 – 0.95	0.13 – 0.38	Up to 3.8	20 - 30	+/-
Li-Ion battery	0.36 – 0.95	0.14 – 0.38	Up to 4.0	35 - 50	+/-
NiMH battery	0.80 – 1.58	0.32 – 0.68	1.25	20 - 40	+
NiCd battery	1.58 – 2.38	0.63 – 1.9'0	1.2	90 - 110	-

From these results it seems to be the most advantageous to use a Lithium-ion-polymer-battery or LiPo battery for short. Apart from these advantages, they also have a small internal resistance and charge quickly and they have very small self-discharge, which makes it possible to store such batteries for a long time without damage. This is particularly useful for suitcases, since they are often not used much, for most people just one per few months. Various applications were weight is an issue also use Li-Po batteries, such as laptops, tablets and mobile phones.

It seems that safety is still an issue with all types of batteries but that these are few differences between the batteries in this case. The most important safety mechanisms that should be present in batteries are mechanisms for preventing over-discharge, short circuit, penetration, over-temperature and charging at the wrong voltage. This means a sturdy and strong casing, as well as coatings and stopping mechanisms. However, there is also a part of responsibility at the user, since the user should take care of the terms of use and the user should not use the wrong charger or operate the battery outside of the recommended temperature range (around -20° to 30°).

According to the table above, a general LiPo battery should cost about €30,-, weigh around 0.60 kg and have a volume of about 0.24 L, which results in a size of about 10 cm x 10 cm x 5 cm. All these three factors are significant but reasonable for the current design.

Wheels

In previous sections, it was said that motorized wheels would be optimal. Now more attention is paid to the mechanism at which the wheel axis turns and how the motor will aid this movement, it has been seen that the wheels itself actually don’t have to be motorized and that this could pose a problem, since the wheels should be able to turn also when the motor is off. Also, previously it was said that omni wheels would probably the most useful to this situation. This also turned out to be not quite true, since the assisting function of the motor works way better and easier when there is just one main axis around which the wheels turn. Furthermore, there is no need for omni wheels in a two-wheeled suitcase, since making turns is easy enough with a two-wheeled suitcase and it is only pulled into one direction.

Taking these points in consideration, it actually seems the most profitable in this case to use wheels similar to wheels in regular suitcases and wheeled luggage. However, there is some things to choose here still. There is some variation in material and most importantly radius of the wheels. It appears that practically all durable suitcase wheels use polyurethane as the material for the wheels. Alternatives are other plastics or rubber, but these materials wear down quicker and have more rolling resistance and are thus found more on cheaper variants of suitcases. Since there is little difference in price and our model of the suitcase is meant to last for a long time, it is a better choice to opt for wheels made of polyurethane. As was said earlier, the wheels will be fastened into the suitcase and will not be separately put underneath the suitcase.

There is actually not much difference in effect between small and big wheels. Small wheels have less surface area and thus less aerodynamic resistance, but this effect is small. Large wheels however have a little bit less rolling resistance (there is less deformation relatively). The idea is that it costs more energy to get bigger wheels at a certain speed than small wheels, but that it costs less energy to maintain this speed for bigger wheels compared to smaller wheels. Bigger effects are that small wheels are of course lighter and occupy less weight and that big wheels can go over rough terrain easier and smoother. Thus, it is not really the speed of the wheels that matter in our case. It is way easier to optimize the speed using an assembly of gears. However, to minimize the weight and size of components in the suitcase is very important in this project, according to user’s requirement R27. The most usual terrain such suitcases will be used anyway is smooth airport floors, so that makes the main advantage of larger wheels is gone. Thus, it seems smart to go for relatively small wheels. Most luggage wheels are about a few centimeters in diameter anyway, the consensus for two-wheeled luggage seems to be wheels with a diameter of 5 to 6.5 cm, weighing 150 to 220 gram, costing around €4 to €7 per set of two.

Total Weight

Table 2:

Component	Weight (kg)
LiPo battery	0.60

Total Cost

Table 3:

Component	Price (Euro)
LiPo battery	30.00

USE Aspects

User Need

The user groups for the above introduced system can be divided into two categories. The primary users consist mainly of any traveller, regardless of the age, gender or ethnicity. However, one could argue that some users are more positively influenced by the system than other, namely pregnant women, the elderly and disabled people. Indeed, pregnant women should avoid lifting any heavy objects . Moreover, the elderly could suffer from back injuries and disabled people have difficulties lifting or moving suitcases. The secondary users are the people that repair such a suitcase, in case of a dysfunction or a broken piece. In this sense, the following needs are important:

Primary users

• Travelers want to minimize the physical effort when carrying a hand luggage

Traveling can be a very tiring process, either through long flights or multiple transfers on top of physically carrying a suitcase. By having an engine-assisting mechanism, the physical effort done by any person is reduced. This allows for a less tiring traveling process especially for the elderly, disabled or pregnant people. The more helpful the suitcase is in assisting, the better it is for the travelers.

• Travelers want to minimize the chances of having their suitcase stolen or lost

Having personal items stolen or lost can be extremely frustrating for anyone, regardless of the value of those items. If the traveler has access to the current location of the items at any time, then the chances of being stolen or lost is reduced. Moreover, in the case of a robbery, knowing that the robber cannot have access to the personal belongings due to being able to lock the suitcase from a distance would be helpful and reassuring whilst the traveler is searching for the suitcase.

• Travelers want to know the weight of their suitcase at all times

In some situations, the weight of the hand luggage can decide whether the suitcase can be taken with the traveler in the plane or should be in the cargo hold. Moreover, the weight is an indication of how much effort the user needs to put when carrying the suitcase. Therefore, accessing such information can be helpful for the traveler.

• Travelers want to have access to all the information stated above

It goes without saying that the traveler wants to access the information such as weight and location in an easy way and at all times.

• Travelers want to afford the system

The system should not be sold at an extremely high price (> 700 euros) so that the users can afford it.

Secondary users

• The secondary user needs the system to be easily maintainable

Maintainability should be taken into account during the design of the system. If the maintainability is low, it costs the user a lot of time, effert and money to fix. Whenever it is difficult to improve the system it is a major disadvantage for both the manufacturer and the user. This will decrease the chances of selling the system because it will generate greater problems in case the system is malfunctioning.

User Impact

The impact of the implementation is manifold. Below the different user segments are listed again for the sake of readability.

Primary users: Travelers

• Less tiring traveling process: this is the result of a robust and well designed engine-assisting mechanism within the suitcase.
• Less stolen or lost luggages: this is a result of having access to the location of the suitcase at all times through the mobile application and the ability to lock the suitcase remotely.
• Better organization and prediction on the traveling process: this is the result of having access to the weight of the suitcase by an internal weight sensor.

Secondary: Maintenance / Supplier

• Increase in demand for their services: this is an outcome of the fact that when more people start using the system, mouth-to-mouth advertising and other indirect forms (social media, press among others) will contribute to increasing product awareness.

Society Need

Similarly to the user category, the society has needs.

• Decrease luggage-related injuries

By assisting in the luggage moving process, vulnerable users such as pregnant women or the elderly have the risk of getting a (back) injury reduced

• Create a better and more pleasant experience for people traveling with carry-on luggage

By assisting in the luggage moving process, traveling becomes less tiring and thus a more pleasant experience. It is believed that there are 8 million people that travel by plane per day . As a result, this means that the experience can be more pleasant for millions of people every single day.

Society Impact

The impact of our solution on the society can be summarized as follows:

• More pleasant travel experience: Automation makes the traveling experience less tiring and more pleasurable
• Less overall injuries: Automation makes dragging the hand-luggage easier and reduces the risk for injuries, which touches thousands of victims every year .

Enterprise Need

Like mentioned before, with the increase of automation in our daily lives, the need to live better whilst also living faster is inevitable. Enterprises play in a large role in this improvement. With regards to the assisting suitcase, travel agencies, airports and bus companies are the ones who could benefit the most. Airports could supply assisting suitcases to frequent fliers, companies or other major customers.

Number of tourists per year (in billion people)

• Automation can lead to more sales

As automation is taking more and more a importance in the world (self-driving cars, autonomous drones to name a few), automated systems are part of the future's technology and users are keen to buy automated systems to make their daily experiences easier and more pleasurable.

• Automating systems can lead to more jobs

The engine-assisting suitcases require design, hardware research and development as well as software research and development. All these fields require employees which are mainly engineers, thus contributing to the job market.

• There are more tourists every year

Since the number of tourists increases every year (as seen in the Figure to the right), companies have more customers and can therefore sell the system to more potential buyers.

Enterprise Impact

The impact of our solution on the enterprise can be summarized as follows:

• More tech companies: Since the system brings more jobs and has the potential to increase the sales, this results in a higher number of companies working on creating and maintaining the system and its components
• More research positions: Since the system can always be improved with new technologies (sensors, software among others), research related to this field will be higher.
• Higher income for power supplying companies: This is the result of having a battery in the engine-assisting suitcase that users can charge with power sockets. This leads to an increase in the number of power sockets in the airports.

Mobile Application Prototype

During the 6th week, the front-end development of the mobile application on Android started. The application was developed using Android Studio.

Future Developments

There are a number of ways in which the technologies discussed here can be extended and improved upon. We will talk about some of these ways here.

Security

One possible feature that could be improved on is security: the suitcase could for example be locked electronically using techniques like facial recognition. This would also be a considerable additional cost, so perhaps an alternative approach could be to use the smartphone app to unlock the suitcase. Most modern phones already have biometric security systems, which could then be used as-is instead of putting this technology in the suitcase itself. However, this would also mean that when a user loses their smartphone, they also lose access to the contents of their suitcase, which is a considerable risk when traveling. It would even be possible that the user puts their phone in the suitcase and locks him/herself out of their own suitcase.

Usage in different contexts

The technology to amplify a user's efforts to help move heavy objects can be extended beyond the use of suitcases. For example, the same methods could help warehouse employees move heavy trolleys. Furthermore, this could prove to be exceptionally useful for wheelchairs, not only making them easier to push from behind, but also helping the person in the wheelchair move by themselves. This could greatly improve the autonomy of the elderly, the disabled, and other groups who are not capable of walking long distances.

Motorised trolleys

In warehouses, large trolleys are a popular way of transporting goods internally. Moving these trolleys around can be a very demanding task and using the kind of assistance discussed here could make this task much easier to perform. It would also be beneficial to safety, since the operators will have more control over the trolley. However, it would still be cheaper, more versatile, and easier to operate than a fully motorised solution such as a forklift.

These warehouse trolleys will most likely carry a much heavier load than the suitcases. As such, the frame needs to be stronger, and the drivetrain should be much more powerful. We can expect trolley to apply much more force compared to the user than in the suitcase, so the control software may need to be adapted to work properly when the trolley is expected to apply almost all of the required force. Furthermore, the trolleys will be in use almost all the time instead of only sporadically, which means that the battery capacity must be as high as possible. They must also be able to charge very quickly, to minimise the down-time due to charging.

Another big difference between the use in suitcases and in warehouse trolleys is that security is not very much of a concern here: the trolleys are only used to transport goods internally and are not at risk of being stolen. Therefore, the unlocking system we described would not be necessary here. Furthermore, there could be many of these trolleys and they may be used by many different employees throughout the course of a day. These employees probably won't want to install an app on their phone and pair it with the trolley every time. As such, we think that the Bluetooth connection and smartphone app should not be used in this context. It would be preferable to link all trolleys to a centralised system where a supervisor can observe their position, battery level, etc. This will also require a different interface: not only should it work on PCs rather than phones, it should also be able to display info on multiple trolleys at the same time. Depending on the specifics of the use case, Bluetooth may or may not be a suitable communication protocol here: after all, Bluetooth is still advantageous due to its popularity and power efficiency, but it is limited in the number of connections that a single device can have, which could be a problem for the centralised observation system mentioned before. Since the trolleys would only be used indoors, Wi-Fi might instead be a viable way of communication.

Trolleys owned by airports

Aside from travelers buying their own electric suitcases, a possibility is to design trolleys that can move multiple suitcases and are property of the airports. This would be useful for families who have a lot of hand luggage, especially families with small children who cannot carry their own luggage. This would have the benefit that all the electronics, batteries, and motors are part of the trolley which is will not board the plane. Therefore, the suitcases can stay simple and lightweight, leaving more weight available for luggage. The trolley, on the other hand, will not have to meet airline regulations, allowing for a larger size, as well as higher battery capacity.

However, this design also has disadvantages: first of all, the airports will attempt to maximise the usage of these trolleys to minimise costs, just like the warehouses. This means that the trolleys will be in use almost constantly, so they have to have a very long range and must be able to charge quickly. Because they should be able to carry many suitcases, their propulsion must also be very powerful. All this combined means that the hardware of the trolley would be much heavier than that of a regular suitcase: the trolley will then be more similar to the warehouse trolley than to our suitcase.

Lastly, this use case changes the way data communication is done: there is now a user, who only move the trolley and expects the trolley to assist in this, and the employees of the airport, who want to know the position and battery level of the trolleys. This means that the travelers will not need to download an app to connect to the trolley, which they probably wouldn't want to do in the first place. It also means that, again, Bluetooth is not suitable for communication, which is now over long distances. Just like in the warehouse use-case, Wi-Fi would be a better communication method here, despite its much higher power demand.

Use in wheelchairs

The last extra use case we will look at is that of a wheelchair. This poses an interesting problem, since wheelchairs can be moved in two different ways: they can either be pushed by someone standing behind them, or the person in the wheelchair can move the wheels with their hands. This means that there will need to be multiple ways of determining the force exerted by the user. However, the rest of the system would be very similar to that of the smart suitcase, so we will not analyse this use case further.

Hardware improvements

Technology is constantly improving, and we can expect some of these advancements to be applicable to this suitcase. For example, with the safety of batteries improving, it is possible that the 100 Watt-hour limit on batteries may increase, which would allow us to improve the range of the suitcase. With batteries' energy density improving, this could be done without increasing the `empty' weight of the suitcase. Other examples of improvements due to technological innovation are more compact or efficient electric motors, lighter and longer-lasting batteries, and faster charging.