R4

Reassuring Rolling Rescue Robot
Back to PRE2018 3 Group8
To the weekly notes

1 Introduction

1.1 Problem statement

On January the 27th, a building in The Hague exploded due to a gas leak. It took the aid workers eight hours to save all residents. Sharwin, who was one of the residents, got stuck underneath his bed. Fireman Arie van Doorijweert mentioned that in the beginning they could only communicate with Sharwin by shouting (1).

The time it takes to localize the victims and to remove the rubble is of the utmost importance for the health of the victims. For aid workers it is necessary to know the specific situation of the victims, so that a personal approach can be made for the rescue. However, often there is no communication in disaster areas between aid workers and victims. This makes it harder for aid workers to estimate the critical situation, because they do not posses enough information to act upon. Victims can be helped quicker and more precise if aid workers would have access to the information of the victims themselves.

However, it is not just the communication between aid worker and victim that should be accomplished. It is also of great importance that the victims are capable of communicating and reasoning, meaning that their state of mind should be calm. This can be quite a challenge in such frightening circumstances.

The rescue robots that are currently in use are solely focussed on performance and its functionality, such as localizing and approaching the victim. However, these engineers have hardly encountered the aesthetics of the robot or the effect on the mental state of the victims. Yet, reassurance has influence on the mental health of post-disaster victims as well as the way in which information can be retrieved from the victims during the event. Therefore, we conduct a research on how a rescue robot should be designed to have a calming effect on victims.

1.2 Project Goal

Our project goal is to conduct research, by means of literature study and interviews, into obtaining a calming effect on victims of a (natural) disaster with a rescue robot. Hereby, we look at the non-verbal communication of a rescue robot (e.g. color, appearance, and approach), as well as accomplishing verbal communication between the aid worker and victim. Based on this research, we will provide guidelines for future designed rescue robots. Additionally, we will design our own rescue robot in order to test if our guidelines are applicable and if people’s opinions match with our literature study.

2 Approach, milestones and deliverables

2.1 Approach

To tackle this project, we started with extensive research on the state of the art. This is done by examining the current literature on the subject. These examined papers outline the current state of the problem, solutions to these problems, and their flaws.

After obtaining a better view on the problem at hand, the USE (user, society and enterprise) aspects are analyzed, to determine why this problem is relevant. These three aspects should always be kept in mind during each stage of the project. These aspects may sometimes ask for different solutions to the same problem, so they must be analyzed to determine which aspect should be taken into account more, and compromises must be made. Also different subproblems may ask for different USE aspects, but two solutions from two subproblems may not always be able to be combined, meaning that choices must be made.

After having analyzed the USE aspects, a scenario will be made where the robot shows of its capabilities, and multiple persona’s will also be made who come into contact with the robot. Research has also been done about the interaction between rescue robots and the human victims. These papers will be used and important factors for the product to have will be determined.

During the whole process, a report will be written in which the process is outlined in a more detailed manner, and which also follows our progress. This report will also more finely describe the problem and the solutions. Because this is done alongside the creation of the actual prototype, planning is important again. Some people will be working on the product and some on the report. This also means clear communication is of utmost importance. Alongside each progress meeting, the group will come together once or twice a week to discuss what has already been done and what should still be worked on. This way it will become clear if the goals will be reached in time and the project is on track.

After all this is done a presentation will be prepared and presented, the wiki will be completed, and all deliverables will be handed in.

2.2 Milestones

• Decide on subject (06/02/19)
• Formulate problem statement (11/02/19)
• Finish literature study (11/03/19)
• Finish sketches of product (13/03/19)
• Finish design of product (18/03/19)
• Present product (01/04/19)
• Finalize the wiki

2.3 Deliverables

• Wiki (report)
• Final presentation
• Research based guidelines
• 3D computer design of the robot

2.4 Planning

PLANNING

Week 1	Literature study (Gialesi, Lotte, Mark, Noor, Romy)		Problem statement (Everybody)		Define users and user needs (Everybody)			Make planning (Noor)		Update Wiki (Every-body)
Week 2			Communication (Gialesi, Noor)	Walking (Mark, Gialesi)	Approach Victim (Romy, Lotte)	Hue of light (Gialesi, Noor)	Looks (Mark, Noor, Romy)	Scenario & persona (Lotte)	Analyze USE aspects (Romy)
Week 3	Scenario and persona (Lotte)					Define requirements and objectives (Everybody)
Week 4	Research face (Noor, Romy)			Research Locomotion (Gialesi, Lotte)			Refine requirements (Mark)
Week 5 (Carnaval)
Week 6	Redefine project goal (Everybody)				Make sketch of the robot (Everybody)
Week 7	Make computer sketch (Mark)		Summarize the research conclusions (Gialesi, Romy)			Explain use of screen (Noor)		Prepare presentation (text and powerpoint) (Lotte)
Week 8	Finalize wiki (Everybody)
Week 9	Presentation					Hand in deliverables

2.5 Who's doing what?

• Lotte Hollander: writing report + wiki, graphic design, prototyping
• Romy Lauwers: writing report + wiki, literature, photoshop
• Mark Wijnands: mechanics, control, 3D design
• Noor Schroen: literature, electronics
• Gialesi Notkamp: literature, electronics

3 User Study

3.1 Users and User Needs

The persona's and scenario based on sources can be found in Appendix A

3.1.1 Primary users

Aid Workers

They search for victims and provide medical help on site. They actually use the robot in the field.

Needs

• Accurate estimated number of victims
• Localize victims more quickly
• Safety while trying to search for survivors/victims
• More data about the situation in critical conditions
• Personal/medical information of the victims
• Easy and fluent communication with victims during the rescue

Victims

Victims of a post-disaster area. They interact with the rescue robot.

Needs

• Less mental issues after the disaster
• Receive medical aid as fast as possible
• Safety
• Communication with a professional aid worker
• Recognize help
• Reassurance of help

3.1.2 Secondary users

Volunteers
Family members, friends, neighbours and others who help with the search.

Needs

• Safety while trying to search for survivors/victims.

Non-profit organisations
Non-profit organisations, e.g. the Red Cross, can give workshops/education about the use of the robot.

Needs

• Safety
• More volunteers

3.1.3 Tertiary users

Government
The government finances the search and rescue.

Needs

• Lower (medical) costs
• Less casualties

Hospitals
Hospitals where the victims are hospitalized. They need to buy the device once but then they will take advantage for longer. They are able to use the device for a long period.

Needs

• Less mental issues after the disaster
• Less casualties
• Quicker discharge of patients

Production companies
It's able for the companies to get more brand awareness when they will produce (a part of) the device. Other less known products of them, that are maybe very good in combination with the device, could get more publicity, which is also a positive impact.

Needs

• More brand awareness

4 Requirements

Requirement	SMART criteria	How?
The solution should enable verbal communication between victim and aid-worker	Specific	The requirement is specific, since it specifies which communication is desired, and between which parties.
	Measurable	This requirement is measurable, since it can be determined if a communication between victim and aid-worker has been established.
	Achievable	This requirement is achievable given its definition. The solution only acts as a means to communicate, and does not imply there necessarily is communication. In some situations it might not be possible for the victim to communicate. The requirement of enabling the communication is achievable though.
	Relevant	Verbal communication is extremely relevant because of two reasons. The first is that it enables the aid-workers to gain more insight in the situation. The physical and mental state of the victim can be determined, as well as getting more insight in the situation itself. A victim can, in some cases, describe what the area looks like, where critical points are and how to approach the situation best. The other important factor of communication has to do with the mental state of the victim. Victims can be calmed down, which is very important. According to Sophie Kruisdijk SOURCE, the possibility of having PTSD after a disaster depends on the kind of injuries, the duration of the rescue and impact of the event. Calm patients are easier to treat, which results in quicker discharge of patients. Furthermore when people cause more damage to themselves by trying to escape, it might result in more difficult medical operations. This will likely result in a longer period of recovery.
The solution should enable aid workers to locate victims more quickly.	Specific	This requirement is specific, since it is clear what is meant. Locating victims is very clear, and whether or not it is quicker can be determined. Therefore it is specific.
	Measurable	This requirement is measurable. Comparing the time to locate victims with the solution to locating them without the solution will give a clear insight in whether it is quicker or not.
	Achievable	This requirement is achievable. Some examples of recue systems can be found that enable aid workers to locate victims quicker
	Relevant	This requirement is relevant, because it enables certain other functions. When victims are located more quickly, they can be helped quicker both in the mental and physical domain.
The solution should reduce the risk of the aid workers	Specific	This requirement is specific because it is clear what the function of the solution should do. The amount of risk can be measured which makes it clear whether or not the risk is reduced.
	Measurable	The requirement is measurable after deployment. The number of injuries of aid-workers when using and not using the solution can be compared. This cannot be measured on beforehand, but an estimate can be made.
	Achievable	This goal is achievable. Several rescue robots have already been implemented, which show a decrease in injury and thus risk for aid-workers. SOURCE
	Relevant	This requirement is relevant because the safety of aid workers is important. Keeping people safe is already relevant. Furthermore, when the risk is lower for aid-workers they might be able to perform their tasks more effectively.
The solution should be easy to operate. Anyone not having used it before should be able to operate it within 5 minutes.	Specific	Easy is different for everyone and thus not specific. The second part makes it specific by saying anyone who hasn’t used the solution before should be able to.
	Measurable	The requirement is measurable, since it can be determined if the device is operated correctly within 5 minutes.
	Achievable	This requirement is achievable. Enough examples of similar systems can be found that are easy to operate. SOURCE
	Relevant	This way the solution can be used by as many people as possible. This makes it so not only aid-workers but also volunteers can use the solution effectively.
The solution should be ready for use. It should be able to perform its functions directly after being deployed.	Specific	This requirement is relatively specific. The functions are not yet mentioned so this might still be vague. These functions however are described by the other requirements.
	Measurable	This requirement is measurable, since it can be determined whether or not the solution works properly directly after being deployed.
	Achievable	This requirement is achievable. Enough examples of similar systems can be found that are immediately ready for use. SOURCE
	Relevant	This way it can be used quickly, without the need of having to set it up. This will increase the speed at which victims are found which could result in a better mental/physical state. Help can be provided more quickly.
The solution should comfort the user non-verbally. The victim should be kept/made calm.	Specific	The requirement is specific, since it is clear what the solution should be able to do. It is also explicitly mentioned in what way the solution should perform its function, namely non-verbally. How to determine whether or not someone is comforted is another question.
	Measurable	This requirement is not as measurable. Comforting someone is hard to measure. Things like heart rate can be measured and when a connection is established the aid-worker can often conclude whether someone is calm or not. This is not a specific thing that can be measured however.
	Achievable	In SOURCE some ways of non-verbal and non-facial affective expressions for appearance-constrained robots are mentioned. These can be used to comfort the user non-verbally which makes this requirement achievable.
	Relevant	This requirement is relevant because, as said before, calming victims down is of utmost importance. Communication will be improved with calm victims. Furthermore victims might be discouraged to escape, which often causes more damage. The non-verbal part is relevant
The solution should be able to fit through a hole with a radius of 10cm.	Specific	This requirement is specific since it clearly specifies what the solution should be able to do. The dimensions are specific as well.
	Measurable	The requirement is measurable since it can be easily determined whether or not the robot is able to fit through the hole.
	Achievable	This requirement is achievable. In SOURCES examples are given of rescue robots that can indeed move through these areas.
	Relevant	This requirement is relevant because it enables other requirements. In urban structures, access to the interior of the rubble pile can be done using three techniques. SOURCE Natural voids formed in a collapse, as well as engineered breaches often begin at the top of the pile. Because the voids are often irregular a pipe is inserted. Therefore it is very relevant for the solution to be able to fit through these pipes. This way it can get to the victims, which enables the solution to find and locate the victim, as well as establishing a connection.
The solution should be able to move in disaster areas. It should be able to climb over obstacles and voids.	Specific	This requirement is less specific since obstacles and voids can be of various sizes.
	Measurable	The requirement is measurable, once the dimensions of obstacles and voids are determined. It can easily be checked whether or not the solution is able to climb over obstacles and voids.
	Achievable	This requirement is achievable. Solutions have already been found that are able to move over certain obstacles or voids. The bigger the obstacles or voids are it can cross, the better, but this is not always achievable, when taking into account other requirements. For the robot to fit through the hole it has a certain maximum volume it can have. This means the voids and obstacles it can climb over are limited as well.
	Relevant	This requirement is relevant because it enables other requirements. For the locating the victim the solution should be able to search the area. To do this it must be able to move through the area. When establishing a connection the solution should be able to get close to the victim. Therefore this requirement is relevant.

These requirements should be completed within a time period of one quartile. The 1st, 6th, 7th, and 8th requirement will be further taken into account because we wanted to focus upon the psychological effect of a natural disaster rescue robot on the victims.

5 Research

Figure 1: Locomotion Brainstorm

Figure 2: Face Brainstorm

During the brainstorm session we discussed possibilities for the design of the robot. We considered the face and locomotion. In the following figures the outcome of the brainstorm can be seen. We eventually decided that the shape is largly depended on the locomotion and face. Therefore we decided to choose five options for the face and locomotion to further conduct literature study. The findings are shown below.

5.1 Face

The following part discusses the external features of the robot. Should the robot have a face? If so, how should that face be executed? To start off, the possible options have been narrowed down to four possibilities. The robot could either have a realistic human- or animal-like face, an unrealistic face, no face at all, or it could have a screen where the face of the aid worker commandeering the robot is shown.

5.1.1 Realistic face

A realistic face is a face that closely resembles that of a human or an animal. An animal face would be preferable to a human face in the case that the body of the robot resembles an animal.

Figure 3: Uncanny Valley

The robot needs to convey empathy, and humans and animals are adept at doing exactly that, for that reason, using a realistic face makes sense. However, there is one problem with trying to use a realistic face, and that can be described using the term ‘uncanny valley’. Uncanny valley is the term used for object that bear close resemblance to humans, but bring a feeling of discomfort with them. This term was invented by robotics professor Masahiro Mori. His original hypothesis states that as the appearance of a robot becomes more human-like, observers' emotional response to the robot becomes increasingly positive and empathetic, until it reaches a point beyond which the response quickly becomes strong revulsion. However, as the robot's appearance continues to become less distinguishable from a human being, the emotional response becomes positive once again and approaches human-to-human empathy levels.

In order to convey the desired empathy using a realistic face, it is necessary that this face doesn’t end up in the uncanny valley. Getting all the features just right, however, would be really hard, and would take a tremendous amount of effort.

5.1.2 Unrealistic face

It would be easier to fabricate a robot with a less human appearance, that would still get a positive emotional response from humans. This is represented by the peak in figure []. This would take considerably less effort, time and money, while still yielding similar results.

5.1.3 Screen/aid worker as face

A study has been done to investigate how robot facial appearance affected perceptions of the robot’s personality, mind and how scary the robot is. Participants rated the robot on these three characteristics. The interaction with the human and robot were done under three different condition in a random order. The robot had a display with either a humanlike face, a silver face or no-face on it.
The results of this study show that the display with the humanlike face was most preferred. Participants state then the robot having most mind, being most human-like, alive, sociable and amiable. The silver face display was least preferred, because it was most spooky, moderate in human-likeness, mind and amiability. The robot with no-face display, so there was nothing displayed on the screen, was rated least sociable and amiable. These results suggest that the more humanlike a healthcare robot’s face display is, the more people are positive about it. Scariness was related to negative impressions. In our case, the display could have the face of the aid-worker on it. However it would be a live view with communication and not a programmed display. This could be positive, because then it is less eerie and people are more inclined to accept help.

5.1.4 No face on display

As can be read above, a no-face display was rated as least sociable and amiable. However there was nothing mentioned about eeriness.

5.1.5 Nothing at all

Without any face the robot would be anonymous. This would influence the perception of the robot. Robots should be more machine-like than human-like according to Oestreicher. They should be easy to use and be safe and reliable. Furthermore in the a study of ROBOCARE Domestic Environment, where the robot interacts with elderly in different scenarios, results showed that the robot without a face was preferred. Also, according to research done by Jamy Li and Mark Chignell on communication of emotion in robots through head and arm movements , head and arm movements suffice to convey emotion.

5.2 Locomotion

The following five ways of robot moving are juxtaposed for their applicability in our design. These options are analysed based on the ease of moving over a post-disaster terrain and the behavioural effect when approaching a victim. The order in which these locomotions are discussed is randomly.

5.2.1 Flying

The complexity of handling uneven terrain means ground robots are tougher to engineer, than robots moving through sea and air. Plane-like flying is also considered as one of the easiest way of moving, because the sky mostly obstacle-free and the step from remote-control plane to flying robots is relatively small. [1] Research on these plane-like flying drones and insect-like flying drones is not developed enough. Because there is too little research and thus too little information about these drones, this locomotion will not be chosen for the design of the rescue robot.

5.2.2 Crawling (serpentine or peristaltic movement)

Serpentine and peristaltic are two alternative ways of robot locomotion that focuses on rectilinear motion.

Serpentine motion is based on the forward movement on a snake which uses ground points in order to push their body forward. The small cross-section and low center of gravity of most biological snakes, coupled with their large repertoire of possible motion sequences, make their bodies very efficient when navigating confined spaces and rough terrains. However, the artificial snakes are generally slow and most (especially those with wheels) are not well suited for travel over rough terrain. This serpentine motion is already being tested by Carnegie Mellon University in a real search and rescue. Even though the robot did not find any victims the Mexican Red Cross workers said that “The robot performed well” and added to that, “they would like to have a similar tool in the future”

Peristaltic crawling is copied from earthworms which propagates a longitudinal wave from the front of the body to the back by varying the thickness and length of its segments. This way, the object can move while keeping a large area in contact with the ground. Therefore, it has the following benefits: 1) The amount of space required by this locomotion mechanism is less than that of others such as bipedal, wheel-based and meandering locomotion, 2) The locomotion mechanism is likely to have stability on irregular ground and inside a narrow pipe. Therefore, it is desirable to apply this mechanism to rescue robots and limited environment exploration robots

5.2.3 Walking

When considering a walking robot for an uneven terrain, the robot should have at least four legs for dynamic stability. By adding more legs, the platform becomes wider and more stable. However, more legs also comes with more moving parts, which require better coordination [1]. Six legged robots are the most stable and popular legged locomotion concept. The walking technique of legged robot is always a challenging task for the researchers. Pre-programmed robots can never successfully interact with the environment autonomously because they have to be reprogrammed for different environments. Hence, a self-learning technique must be incorporated to make the robot learn things as per its interaction with the outside environment. [2]

5.2.4 Flipping

Body-flipping locomotion is achieved by active flipping of the robot’s body. The benefit of this type of motion is that it is fast and the movement can be predicted efficiently. However, flipping locomotion in robotics has so far only be tested on stairs and not in post-disaster areas. Also, the robot needs more space while flipping than a robot has more contact with the ground. Therefore, it will probably not be suitable as a rescue robot.

5.2.5 Rolling (wheels)

Rolling is the easiest and most common way of robot locomotion, e.g. cars and iRobot [1]. The main reasons for this are their easy controlling, stability and simple mechanical design. Wheeled locomotion is preferred over other modes of locomotion due to their power efficiency, faster running capability but traction on rough terrain, its control and stability are their primes areas of concern. For static and dynamic stability, the robot should have three wheels. Controllability of wheels becomes harder when the maneuverability of the wheels increases. This is because an small error (e.g. change in speed) can cause a robot to deviate from its path and make it difficult to control. However, maneuverability is important for a multi-terrain robot in order to move swiftly in any direction on the ground. Omnidirectional powered wheels are used which enhance the maneuverability in multi-terrain environments [2].

5.3 Light

When the robot moves towards the patient in the dark environment under the rubble, this can be quite stressful for the patient. The robot would most likely not be completely silent when moving. To make the patient ready for what approaches him/her the robot could emit a light, making its form visible. Additionally, when considering the requirement about non-verbally calming the victim, it is proven that light can influence the emotional state of a person. As shown by multiple researches, the colour green and blue are mostly associated with the emotions peaceful and calming. If the robot would emit a light in one of these colours this could already help the patient to calm down. When it is dark, traffic light green is the most visible colour. As green is better seen in the dark, the victims could spot the robot better and the feeling of reassurance would get bigger, as they know help is on the way.

When the robot approaches victims, it should not discomfort them by exposing to light that is too bright. Thus the robot would have to send a light with a certain lumen level. Lumen is the unit of measure for the brightness or intensity of light. Brightness under 3500 lumens is still acceptable, but when the light reaches over 4000 lumens the light will get to bright for our eyes to handle. People will instinctively narrow their eyes when exposed directly to this much lumen of light. At a strength of 10,000 lumens the light can cause permanent damage, when shone directly into someone’s eyes for a long time.

5.4 Approach of Person

The psychological study of how people establish mutual distances in communication is called proxemics. However, this study might not be applicable to robot-human communication. When a robot approaches a human, several considerations should be encountered in the way of approaching. (1) Personal Space: a social convention that defines a region of space around individuals as personal, (2) Direction of Encounter: the preferred direction from which a person would like to be addressed, (3) Feasibility of interaction: the region of space where interaction is feasible.

In the thesis ‘Approaching Independent Living with Robots’ by Elena Torta, these three positions are tested and evaluated for a sitting and standing positions of the test subjects. The results concluded a best mutual distance between 140-200 cm. Also, the central direction of encounter was preferred most. The experiments were conducted with the Nao robot and only encounter sitting and standing positions. In our scenario it is also possible that the victims are lying down and have less freedom of movement, which influences the prefered approach distance. Also, the appearance of the robot is different.

Furthermore, an important thing to consider is the speed at which the robot approaches the victim. In an article about non-facial and non-verbal affective expressions for appearance-constrained robots (26) the following test is done. A robot is used in a dark, high-fidelity, confined-space simulated disaster site in two different modes. There is a standard and an emotive mode. In the emotive mode the robot exaggerates his movement. As the robots approached the participant they would exhibit cautious and interested behaviors through a creeping movement, and would slowly raise similar to a dog or squirrel investigating something unknown. When really close (in the so called intimate zone) the robots movements are very limited and controlled so the participant isn’t frightened. Furthermore, after initial contact with the participant, the robot would slowly back away from the participant showing concern and attentiveness. The study showed statistically significant results which indicated that participants felt the robots were more calming, friendly, and attentive in the emotive mode. This improved the social human-robot interaction.

5.5 Audio

In order to establish verbal communication between victim and aid worker, audio is needed. On the robot this will consist of a microphone and a speaker which are both wirelessly connected to a unit at another place where the aid worker is. So, the audio system will be almost like a phone connection between the victim and the aid worker. To save power, these are remotely turned on and off by the aid worker. There has been looked into which kind of communication is comfortable for the users.

5.5.1 Submarines

5.5.1.1 AT (Acoustic Transmission)

This audio transmission works better underwater, because of the way that sound waves travel underwater. Therefore it will not have the same effect above water. Thus it is not the type of communication for our robot.

5.5.1.2 VLF (Very Low Frequency)

“Due to the low frequency a VLF broadcast antenna needs to be quite large. In fact, broadcasting sites are usually a few square kilometres.” As the broadcasting sites are so big, this is not a useful way of communication.

5.5.1.3 ELF (Extreme Low Frequency)

As is stated on the wikipedia site: “Due to the technical difficulty of building an ELF transmitter, the U.S., Russia, and India are the only nations known to have constructed ELF communication facilities.“ This means that the ELF communication is not yet usable for the robot and therefore it can perhaps be used in the future, but not yet.

5.5.1.4 SRT (Standard Radio Frequency)

This is the radio that the submarines uses when above water. This is because it does not work underwater. Thus it will also not work when the signal has to go through debris.

5.5.1.5 CART (Combining Acoustic and Radio Transmissions)

Just like AT, this is based on the principle that sound travels in another way underwater than above water. Thus it will not have the desired effect in our scenario.

5.5.1.6 UM (Underwater Modems)

Just as the AT, this technique makes use of the fact that sound travels in a different way underwater as it does above water. Therefore this method is not good enough.

TTE (Through-The-Earth) communication systems uses ultra-low frequency (300–3000 Hz) signals that can penetrate through several hundred feet of rocks and such. It is used in mining operations, especially in disaster situations where the standing communication systems don’t work anymore. It works by placing an antenna cable in a loop along the perimeter of the area where coverage is needed, and the transmission propagate through the rock, using them as a medium to carry the signals. Currently, there are a couple of (handheld) devices that are capable of using this technology, one of them being the Personal Emergency Device (PED), which has functionalities close to what is desired in the R4 . The PED allows for two-way audio and data communication. This shows that the technology for the communication we want, fits inside the robot. At this moment, there is no device that supports video on this frequency, but that seems to be due to the fact that that has not been necessary up until this point and not because it’s not possible. TTE should be the answer we are looking for.

The victim will talk with an aid worker and not with the robot itself because programming a robot to include speech recognition and speech is extremely hard. Also, with the current state of the art, robot voices are pretty monotone. They don’t convey a lot of emotion or empathy, which is desirable in cases where the robot will be working. Of course, we could work on making a non-robotic sounding voice for the robot, but this is not the goal of this project. A human however, can convey empathy even with only their voice. This is confirmed by the fact that for a really long time, and still, psychoanalytic sessions take place over the phone for some people due to personal circumstances (35). According to Leffert (36), he had nine analyses in which he found that telephone sessions were indistinguishable from in‐person sessions, but he did not support his findings with detailed process. (Psycho)therapists must have a deep understanding of empathy to be able to talk with their patients. If telephone sessions can be as effective as in-person sessions, that means that therapists (and other people), can also convey a certain level of emotion and empathy over the phone, which is needed to calm down victims.

6 Results Literature Study

At first research has been done to the non-verbal aspects, namely face, locomotion, colour of light and the approach. A robot having a display with the face of the aid worker on it was deemed to be the best option. On the display a live view of the aid worker will be shown, because that is less eerie than a preprogrammed face for the user. Furthermore a study has shown that a robot with a face seems to be more human-like and sociable. This can be considered as an import aspect, because of the non-verbal requirement. When considering the suitability of robot motion in disaster areas and its size limitations, rolling is the most fitting design choice. This locomotion fits the requirement that it should be able to climb over obstacles and voids. Furthermore a robot with this locomotion is easy to control and has high maneuverability and dynamic stability. We will use three wheels for the robot, because it has the highest maneuverability. We will use the color green for the light that the robot would emit. The light that the robot emits will by below 4000 lumens at all times, as it would otherwise cause discomfort to the victim. Therefore the standard of 3000 lumens is chosen as standard light emission.

In the thesis of Elena Torta the experiments were conducted with the Nao robot and only encounter sitting and standing positions. The results concluded a best mutual distance between 140-200 cm. However, in our scenario it is also possible that the victims are lying down and have less freedom of movement. Also, the appearance of the robot is different. Adding to that, the robot might not have the freedom to move that the Nao robot had, as there can be debris in the way. The other study where the robot is used in a dark, high-fidelity, confined-space simulated disaster site in two different modes, showed statistically significant results. They indicated that participants felt the robots were more calming, friendly and attentive in the emotive mode. This improved the social human-robot interaction.

At last communication and in specific the audio is investigated. As it is very hard to program the robot with speech recognition and robot voices are still very monotonous, the best choice is to let the aid worker speak to the victim. The aid worker can convey emotion and empathy which is hard to near impossible for an automated voice. So the robot will consist of a microphone and a speaker. Furthermore, if telephone sessions can be as effective as in-person sessions, that means that therapists (and other people), can also convey a certain level of emotion and empathy over the phone, which is needed to calm down victims.

[nog wat voor communicatie]

7 Design

To get a more clear overview of the conclusions of the research a design was made. The aim of the design is to show how each aspect can be implemented in the rescue robot branch. The sketch of the design showing the different components can be seen below. The final 3D design can be seen in figure 2 and 3:

Figure 1: Robot Design Sketch

7.1 Body

A first thing to notice is how the robot has round surfaces and no sharp edges around the body. From both the literature and our own study, it was concluded that rounder edges convey a safer feeling than sharp or pointy edges or pieces. The body is made bigger at the front, to make sure the screen could be fitted. Another important thing that is not directly visible in this figure is the light. Inside the body of the robot a green light is fitted. By making use of transparent materials the light can be emitted through the whole body.

7.2 Wheels

As said in the last chapter wheels make for the best maneuverability and dynamic stability. The design makes use of two big wheels in front. These create stable and easy movement. Steering can be done easily by making one wheel rotate faster than the other one. On the side of the wheels a strip of green light has been added. Because the light normally comes from the body, but the wheels block part of the view when looking to the robot from the side, the light of the body itself is less visible. The strip of light on the side of the wheels makes it so the robot is also easy to spot from a side view. For stability the back wheels have been added. These are smaller and only have a purpose in keeping the robot straight.

7.3 Height adjusters

Another important aspect of the robot is the approach. Both how close the robots gets and the speed at which it moves are not of importance to the visuals of the robot. One of the things mentioned in the approach however, is the way it moves. A robot in the aforementioned emotive mode showed that social human-robot interaction is improved by making a creeping movement at first, after which they would slowly raise similar to a dog or squirrel investigating something. After initial contact the robot should slowly back away showing concern and attentiveness. To do this the height adjusters have been added. The body of the robot can be moved up and down. This way the robot is able to do things like slowly raising up.

7.4 Screen

Another important decision was to not use a human-like face. For the communication and audio a screen with the aid-worker was chosen. Therefore a screen was added. The screen can be moved outwards, using a mechanical arm on the back of the screen. This will make it so the screen sticks out further, and can be tilted in different directions. This will make it easier for the victim to see the aid worker, even when movement for the robot or the victim is restricted. The robot will not have to move its whole body, only the screen.

7.5 Sensors

Last of all the sensors are added. These are of course needed to measure the world around the robot, so it can move through the environment the right way. Which sensors are used is not defined specifically, since this is not the scope of this project. The placement of these sensors influences the visuals of the robot quite a bit as well. When placing them above the screen they might look like eyes, which might in turn be creepy or may cause discomfort. Therefore they have been placed under the screen. Their placement now makes them look like a small mouth. Together with the screen this makes for a face, but a very unrealistic one. This has purposely been done to avoid the ‘uncanny valley’ problem that has been mentioned before.

Figure 2: Robot Design Front View

Figure 3: Robot Design Side View

uitleg Methodology of het interview + type vragen die we willen stellen

8 Questionnaire

8.1 Methodology

The questionnaire was made to find out whether the literature study coincides with the reality. All the different aspects of the literature study were taken together in multiple designs. The popularity of these designs was tested in the questionnaire. The questionnaire asks the participants keep the emergency scenario in mind while filling in the questionnaire. The scenario is sketched is a natural disaster in which the participant ended up. The next three questions ask the participant to attach emotional values to three different robots. Two of the robots are rescue robots and the other is a robot which is round.

Next, the partakers were asked to compare two designs. This was to see the difference - if any - between a screen that has pointy edges and one with the edges rounded off. Another subtle difference were the sensors that were visible placed under the screen to form a face-like appearance. Lastly, different colors were shown to see if the green/blue preference of the literature study came back when the color was implemented in the design or a random robot.

8.2 Results

9 Discussion and Conclusion

In a lot of ways the answers of the questionnaire coincide with the results from our literature studies. There are also some unexpected outcomes, which will be discussed in this segment.

The first unexpected outcome was the major preference for the color blue. The expected result was that green and blue would lie close together, but when looking at the preferred color of the Nao robot blue had a vast majority (48,1%) with respect to the 11,1% of people that voted for green. The results for the R4 were a bit more in line with the expectations though, with 40% choosing the blue R4 and 33,3% choosing the green one. Green was also referred to as looking eco-friendly, which is not the goal of the R4. People also associated blue more with cleanliness, caring and calmness, which are the desired attributes of the R4. This result, where the blue robot was preferred over the green one, did not line up with our conclusion. Research showed a slight preference for green because green light is easier to see. However, since the choice for green and blue was so close in our research, and people outed a strong preference for blue, the R4 is blue, and not green, as it originally was.

There was another result that was not unsurprising, but it did give us the affirmation that we had made a correct choice regarding the shape and general appearance of the robot. A spider robot was mostly perceived to be scary, creepy, and technical/mechanical. These descriptions do not match with the goal that was set for this project. The robot needs to sooth and calm a victim, not creep out the victim. Two other pictures of robots were shown in the questionnaire, one with a snake-like robot and one with a similar shape to the R4. The snake-like robot wasn't received very well either, but better than the spider, being called mostly snake-like, creepy, flexible. The round robot got more positive answers like friendly and futuristic. This is not exactly the desired answer, but it sounds better than the descriptions for the snake and spider.

Figure 5: Final design front view

Figure 6: Final design back view

10 Reflection

In this chapter a reflection will be given about the final project. The most important choices will be mentioned, as well as why these choices were made, and how they influenced the final result.

One of the biggest choices of the project was the choice to abandon the first concept we had. From the first day we had an idea to create a spider robot for rescue operations. This however was not well argumented by literature or anything else. After having set up the user needs and consequently the requirements, it became clear the spider robot might not be the best solution for our problem after all. This also partly had to do with the goal we set. At the start of the project the idea was to create a rescue spider robot, to enable communication between victim and aid worker. However, when it became clear we had to be more specific and create something new, this seemed to be a hard task. Spider robots already exist, and to make a better, faster or smarter one seemed impossible in the given timeframe. Therefore we had to start looking for other ways we could make an improvement. This meant that the goal of the project changed, which also meant different requirements and a different end product. When the goal changed to improving non-verbal communication the spider idea had to be looked at critically again. When looking for the best solutions for each requirement and combining them, it turned out that the spider robot was not the best fit anymore. Therefore this idea was not worked out further. This did mean a lot of the work we did before this point was lost. We could still use some of the written parts, but most parts about the robot itself had to be removed.

The conclusion that can be drawn from this is that it is important not to start with a too specific idea. We started with a final concept without looking at why this design was chosen. This was improved when the goal changed though. We learned from the first part and started by setting up user needs and requirements, from which the design followed. The effect this had on the final result was that no actual design was made, only a computer design. The choice to change the goal of the project was made too late for us to physically create the concept. This was however not the goal anymore, since the results are meant to be guidelines when creating or designing rescue robots, not a single model that is better than the rest.

Another point to look at critically is the division of labor between group members. We agreed to meet twice a week for two hours each. At these meetings we would sit together and discuss what needed to be done. The problem however was that at the start of the project these tasks were often worked on together, instead of dividing them between the group members. This caused the communication to be suboptimal. Later on in the project this was improved. Tasks were assigned individually, and a lot more progress could be made. This also meant the meetings were more productive. There was a lot more to talk about content-wise, which meant progress was made faster. This was very beneficial for the final result.

Appendix A

Sarah Janssen

Photo 1: Sarah Janssen

Name Sarah Janssen
Age 38
Work chief fire officer
Family married and two children, Jimmy (5) and Lisa (7)

Sarah is a devoted wife and mother, and works forty hours per week at the fire department. Besides combating fires and helping evacuees to safety, her job is to oversee the delivery of emergency services to accident scenes and burn sites. She is continuously working on improving the emergency services delivery for a quicker response time and rescue.

Frustrations

• Estimation of the number of victims is not accurate
• Evaluation of the victims’ personal situation in a disaster area is based on limited information
• Communicating with victims during the rescue is difficult

Goal
A way of communicating with the victims in a post-disaster area to gain information about their health and position for a more efficient and safe rescue.

Steve Clinton

Photo 2: Steve Clinton

Name Steve Clinton
Age 53
Work real estate investor
Town Tampa, Florida, USA

Steve lives together with his wife Daisy and his 7-year-old son William. His family evacuated to their family in South Carolina when the Hurricane Irma arrived in 2017. Steve stayed home, which got destroyed, and he got stuck under the rubble. Luckily, professional aid workers saved Steve after 10 hours and Steve survived the natural disaster. While looking back at the event, there are some things that he would have done differently if he could.

Frustrations

• The feeling of helplessness, because he could not contribute to his rescue
• The feeling of anxiousness, because he did not know when he would be found and how long the rescue would take
• Fear that the event might repeat itself

Scenario

An earthquake of Richter magnitude 5 has caused the collapsing of a building. It is estimated that at the time of the collision, six people were inside. Sarah coordinates the rescue teams and rapidly has to decide how to rescue these people. With the help of R4, she is able to virtually reach the victims and communicate on their personal situations. She can calm the victim by telling them help is on their way and analyse their situation based on the information provided. This way, Sarah knows the health conditions of the victims and how to remove the rubble, and can therefore bring everyone in safety more efficiently.

Appendix B

In a post-disaster scenario, it is of utmost importance to get a calming response from humans interacting with robots. This paragraph looks at whether the way a robot moves, the postures it displays, and its orientation can make a significant difference in how humans respond to the robot.

In most cases it is clear the direction from which the robot approaches the human is important. How it moves is often not mentioned though. There are several studies done about the direction, but most of these have been done with a single robot. This makes the question a lot harder. Furthermore the question is quite likely influenced by the victims personality, as is the case with approach distances (28).

One important thing to consider is the speed at which the robot approaches the victim. In an article about non-facial and non-verbal affective expressions for appearance-constrained robots (26) the following test is done. A robot is used in a dark, high-fidelity, confined-space simulated disaster site in two different modes. There is a standard and an emotive mode. In the emotive mode the robot exaggerates his movement. As the robots approached the participant they would exhibit cautious and interested behaviors through a creeping movement, and would slowly raise similar to a dog or squirrel investigating something unknown. When really close (in the so called intimate zone) the robots movements are very limited and controlled so the participant isn’t frightened. Furthermore, after initial contact with the participant, the robot would slowly back away from the participant showing concern and attentiveness. The study showed statistically significant results which indicated that participants felt the robots were more calming, friendly, and attentive in the emotive mode. This improved the social human-robot interaction.

Figure 1

After looking at the best way to approach humans, the actual technical difficulties of reaching these solutions should be looked at. What different types of walking mechanisms are there, and which one suits the found solution best. Up until this point it was taken for granted a spider robot would be chosen, but is this the best possible idea? In a paper about the design of a spider robot it is mentioned that six-legged robots present opportunities by having a small size and practical mobility. The number of legs provide more controlled balance when comparing them to the majority of multi-legged robots. In the rescue application they could be very beneficial. Furthermore multi-legged robots are more versatile than wheeled robots, and can traverse many different terrains. The problem is the complexity of the robot, as well as power consumption. For this problem however, the versatility and ability to move through many different terrains is of more importance than the complexity and power consumption.

Figure 2

One of the ways to make the spider robot walk is by looking at an actual spider and its joints. At each joint a servo is placed so the root resembles a spider as much as possible. The design is shown in Figure 1. This will make the robot able to move in every possible direction. This design has a few problems though. The first problem is the difficulty. Since three servos are used per leg, the system has a lot of degrees of freedom. Even standing still is a difficult thing to set. Another problem is the amount of weight the robot can carry. Since the parts are connected via servo’s the weight load is determined by either the strength of the material or the maximum moment about the servo axis. There should be at least two servo’s, something to control the servos, a power supply and an interface next to the servo acting as a moment on the servo’s of the legs. This either requires very low weight parts or very strong servo’s. Another option to reduce the moment is shortening the rod connecting servo b and c. This however will make the robot way less versatile, since smaller steps can be taken.

Another way of making the robot move is making use of Klann’s mechanism. This is a planar leg mechanism that uses a single rotary motion. There is one central crank that can be rotated using a rotary actuator such as an electric motor. All other links and pin joints are unactuated and move because of the motion imparted by the crank. The position and orientation of each of these links is uniquely defined by specifying the crank angle. Therefore the mechanism has only one degree of freedom. The mechanism can be seen in Figure 2.

Another way to make the robot move, similar to Klann’s mechanism is the so called Jansen's linkage, developed by a Dutch kinetic sculptor called The Jansen. This mechanism makes for a simple simulation of an organic walking motion using only one rotary input.Again, the position and orientation of each of these links is uniquely defined by specifying the crank angle. Therefore this mechanism also only has one degree of freedom. The mechanism can be seen in Figure 3.

Figure 3

When comparing these three options a lot of things have to be taken into account. How smooth is the movement? What load can it hold? How hard is it to build? What is the construction expense? Can it run through rough and difficult environments? How hard is it to control? How much maintenance does it need?

To come to a conclusion the robot should move in a specific way. First of all the robot should not move too fast. The robot should look cautious at first, starting low at the ground and taking careful steps. After a while it should raise. In the intimate zone the robots movement should be very limited and controlled. The mechanism used to walk will be the Jansen’s linkage. This linkage has most of the benefits of Klann’s mechanism, but also makes for a more smooth and especially better looking movement. The leg looks more stable and sturdy than the Klann mechanism.

References

[1] Newspaper article on gasexplosion in The Hague

[2] Scientific bout the approach of the victim <- Is much 'goeie'

State-of-the-Art

Spider Robot and Motion
[3] This paper looks at certain safe points where the spider robot can place its feet and where not in a plane.
[4] This paper looks at certain points where the spider can and cannot place its feet.
[5] This paper looks at a spider robot that climbs autonomously in pipelines. Could be useful for the small spaces.
[6] This paper is about the capabilities of the spider and studies the foot force and torque distribution of the spider in different conditions and compares the leg configurations in order to minimize the torque effort.
[7] This paper discusses foot designs and fabrication for use with a spider-inspired climbing robot.
[8] This paper is about a four-legged spider robot that learns how to move in its environment and reacts to physical changes.
[9] This article discusses a dragline-forming robot inspired by spiders
[10] This is the site of Robugtix. This company has a small spider robot, which can make smooth, life-like motions. The toy comes equipped with a 3D printed body, 26 motors, and microcontroller board pre-loaded with the Bigfoot™ Inverse Kinematics Engine.
[11] This paper focusses on a spider-imitated robot used for rescue

Relevant Rescue Robots
[12] This paper is about a rescue robot with debris opening function
[13] This paper is about MOIRA the Mobile Inspection robot for Rescue Activities.
[14] This paper is about a robot that can move the debris.
[15] This patent is about an all-terrain rescue and disaster-relief robot.
[16] This paper discusses an aerial search and rescue robot and its application to a specific earthquake.
[17] This paper is about modular, reconfigurable rescue robots.
[18] This article describes a so called WALK-MAN robot in post-earthquake scenario's.
[19] This patent is for an autonomous detection system and method of rescue robot in disaster area for complex environment
[20] This patent is for a full topography intelligence rescue robot with self-balancing objective table
[21] This patent is about an emergency relief goods transporting robot
[22] This article presents several different types of robots that can be easily deployed in rescue operations
[23] This chapter summarizes the status of rescue robotics

Disaster Rescue
[24] This patent focusses on a method for priority evaluation for robots under disaster rescue environment
[25] This article is about a challenge that aims to accelerate the development of robots that can help humans, not only with nuclear emergencies but also with fires, floods, earthquakes, chemical spills, and other kinds of natural and man-made disasters.
[26] This patent is about video search and a rescue robot based on ZigBee wireless positioning and search and rescue method
[27] This article discusses a simulation project for disaster rescue
[28] This article discusses disaster robotics and different ways of reaching victims.

Rescue robot interaction
[29] This paper provides a short tutorial on how robots are currently used in urban search and rescue and discusses some robot-human interaction issues encountered over the past eight years.
[30] This paper talks about non-facial and non-verbal affective expressions for appearance-constrained robots.
[31] This paper presents findings from field trials observing human-robot interaction between certified rescue workers and two types of tactical mobile robots at a rescue training site.
[32] In this article the influence of personal traits of a person on the approach distances of robots is discussed.

Prototypes of a spider robot
[33] This paper is about a high tech spider prototype, mady by reseachers of the Fraunhofer Institute for Manufacturing Engineering and Automation IPA. The prototype will provide emergency responders with an image of the situation on the ground, along with any data about poisonous substances. Future plans envision its use as an exploratory tool in environments that are too hazardous for humans, or too difficult to get to. Furthermore, the prototype is very cheap to produce.
[34] This site has an instruction guide to print 3D parts of a spider robot with four legs. It is possible to place an arduino in the middle of the design.
[35] This patent is for a novel rescue robot that can efficiently walk in a complex post-disaster area through a design of a spider-like structure

Robots with interesting factors
[36] This robot is portable and foldable and quickly carried in a backpack to a site where inspection, exploration, search and rescue, and other tasks are required to performed.

Regulation
[37] Regulation and entrainment in human-robot interaction

Communication
[38] Clinical issues in analyses over the telephone and the internet
[39] Analysis and Psychotherapy by Telephone: Twenty Years of Clinical Experience