Robotics is prominent in industries such as manufacturing, military, security, and space exploration (GeeksforGeeks). Robotics is becoming more powerful today due to advances in artificial intelligence, also known as AI. As a result, there’s been much concern manifested in entertainment and media about robots becoming dangerous in the future from AI. These are important and valid concerns, but for this project, I’m going to focus on the technical benefits of coordinating robots. However, those concerns should receive their due diligence when work like this is deployed.
The coordination and collaboration of robots aims to achieve better results than what the one robot could do. This project focuses on using game theory to do that. One article describes how it works: “the goal of the game is for the team to find a coordinated solution that will maximise the rewards for each robot and the total reward of the whole team” (Chu). In one research article, game theory is applied to multiple robotics searching (Meng). I solve a scenario down below to illustrate how this process could work. The problem is modeled as a matrix form game where the players each choose various strategies and receive payoffs from them.
The scenario I’ll model for this project is as follows: Two robots cleaning a house. The two robots are both motivated (by a reward) to clean as much as possible by themselves. The scenario involves needing to clean two rooms and the robots each choose to try and clean the left or right.
Players — Robot 1 and Robot 2
Payoffs (Cardinal Ranking)
2 — AB, BA
1 — AA, BB
2 — AB, BA
1 — AA, BB
AB and BA result in (2, 2) because the robots are working in separate rooms and not conflicting/bumping into each other and therefore being more efficient.
AA and BB result in (1, 1) because the robots waste time bumping into each other/conflict
*Note (mistake) — should be 1B in the bottom left
The Nash Equilibrium in this scenario is AB and BA for a payoff of (2,2). Nash equilibrium means that neither player should change their strategy if the game ends up in this situation because it would result in them having a worse outcome. As a result, the robot should decide to move left if the other moves right, and vice versa. The outcome is also Pareto optimal because there is no better outcome for either player/robot.
If the robots didn’t have game theory applied, they may have entered the same room, experienced conflict, and achieved a suboptimal outcome. Finally, while my project focused on robot-robot collaboration & coordination, another impactful area involving game theory is human-robot collaboration (in this case, iterative learning is applied instead of a matrix model). In particular, manufacturing can be improved through that collaboration: “Manufacturing processes are faster, more efficient, and more cost-effective when humans and robots work together” (Sobalvarro). Overall, I hope working through the scenario of applying game theory to robotic coordination shows the versatility of game theory and some problems in robotics. There’s still so much more about robotics as well as game theory than what was talked about in this project!
Thank you for reading through my project! I would definitely appreciate comments on how this project could be improved since they’d make it better. Or, I’d love to hear about another interesting example of robotic coordination, or just some interesting applications of robotics. Thanks!
Chu, Hongyang. “Game Theory.” FutureLearn, www.futurelearn.com/info/courses/robotic-future/0/steps/26370.
GeeksforGeeks. “Top 10 Applications of Robotics in 2020.” GeeksforGeeks, 3 Nov. 2020, www.geeksforgeeks.org/top-10-applications-of-robotics-in-2020/.
Meng, Yan. “Multi-Robot Searching Using Game-Theory Based Approach – Yan Meng, 2008.” SAGE Journals, journals.sagepub.com/doi/full/10.5772/6232.
Sobalvarro, Patrick. “Here’s Why Human-Robot Collaboration Is the Future of Manufacturing.” World Economic Forum, www.weforum.org/agenda/2020/08/here-s-how-robots-can-help-us-confront-covid/.