AutoRef MSD 2019

Internal Navigation

1. Main

1.1 Background

1.2 System Objective

1.3 Project Scope

1.4 Team

2. System Architecture

2.1 Architectural Framework

2.2 System Decomposition

2.3 Layered System Hierarchy

2.4 References

3. Implementation

3.1 Getting started with Crazyflie 2.X

3.2 Localization

3.3 Path Planning

3.4 Visualization

3.5 Vision System

3.6 Integration

4. Conclusions & Recommendations

5. Manuals

5.1 Visualizer

5.2 Simulator

5.3 Useful Links

NOTE: It is important to bring to light that due to global pandemic of COVID-19 caused by SARS-CoV-2, the project ended prematurely before all the objectives could be met properly. The whole team of MSD 2019 would like to express our deepest sympathies to all those who succumbed to this pandemic and their family members. We also are indebted by the services provided by doctors, nurses and other health care professionals all over the world who have worked tirelessly to keep this pandemic under control.

This wiki page has been divided into several sections as shown in the Internal Navigation pane on the right to make things easier to understand and for readability purposes. Section 2 discusses about the objectives and scope of the project in detail. Next, Section 3 presents the system architecture developed as a part of this project. Furthermore, Section 4 discusses the different implementation aspects of the developed solution. Section 5 presents the final concluding remarks and provides recommendations for the future. Finally, manuals describing set up of different software and hardware components of the project are included in Section 6 of the wiki page.

Background

Football is a long-standing cross-generational passion and represents a growing industry with a market size of several billions of euros. As a result, technology is being increasingly used in football, ranging from automatic goal detection and use of trackers to monitor players’ performance to soccer robots where the players are robots themselves. In a 2014 interview, Franz Beckenbauer, honorary Bayern Munich president and former famous player, predicted that in the future drones will replace human referees.

System Objective

An autonomous referee may provide several advantages with respect to a human referee. First, human referees, naturally prone to human errors, are one the main sources of controversy in the game; they have their own interpretation of the rules, introducing a non-predictable factor often leading to unfair situations in a game where both financial and emotional stakes are high. An autonomous system would mitigate this, and in particular remove the unfairness factor - every game would be refereed according to the same algorithm.

The use of an autonomous referee also makes sense in a robot soccer match, where, although the players are robots, the referee is still human. In particular, some rules are rather difficult to check and enforce by a human referee. For example, when a free kick is given to one of the teams, "all other players of the free-kick awarded team can stay anywhere on the field except in a circle with a radius of 2m around the ball until the ball is in play" and "all players of the defending team can stay anywhere on the field except in a circle with a radius of 3m around the ball until the ball is in play. One robot may stay anywhere inside the penalty area (except goal area) of its own team, even if the distance to the ball is shorter than 3m". Checking if these rules are exactly met by the robots is typically hard for a human, but rather easy for an autonomous robot referee with for instance a vision system.

Project Scope

While a camera-based autonomous systems covering virtually any possible game situation can provide a solution for professional games at major leagues and stadiums, it is rather expensive in many other contexts such as regional games; for example in the World Cup 2014, the production plan used a high-end system with 34 cameras. A moving camera provided by the drone can therefore replace the need for such an expensive solution, and therefore appeal to a large market.

The technology to automatically enforce the rules of the game based on video is not available, a camera system capturing important game situations can assist a remote auxiliary referee, to which it provides video and repetitions in real-time. In turn, the remote auxiliary referee informs the main referee of his/her decision. This has been introduced recently in several major European (e.g. in The Netherlands) and United States leagues.

From a broad perspective the goal of the present project is to contribute to this vision and create an autonomous robot referee system using drones. The drone can be used to provide images to a remote referee, who decides based on images and repetitions. The remote referee can then inform the drone or another onsite referee system about decisions (which in turn are communicated to the teams). In the (perhaps distant) future, the drone will automatically and autonomously make decisions.

Team

This project was carried out in the second module of the 2019 MSD PDEng program. The team consisted of the following members:

• Ankita Kalra
• Harun Salman
• Karl J. Blacker
• Pranjal Biswas
• Song Guo
• Tamoor Ali
• Wan-Yi Tang
• Yusuf Salman