IA for Robotics: Rhoban project

One of the challenges faced by the Rhoban project is to develop software and methodologies for computing strategies for teams of mobile agents cooperating and/or competing with each other. In particular in the context of the “Robocup” competition, for example in the simulation, small size and humanoid leagues.

In this context, a team of robots is competing against another team in some physical environment.
The control of the team may be centralized (e.g. in the small-size league of Robocup) or distributed (in the simulation-league). A problem-solver has been developed in LaBRI by the Rhoban team (PhD thesis of Ludovic Hofer). It combines several reinforcement learning techniques (RL) for computing good strategies in Markov decision processes. These strategies are precomputed and represented by regression forests,
which allows a mobile robot to perform real-time decision making at a very low computational cost.
For the moment, this approach scales up to computing cooperation strategies between a small number of robots, but not an entire strategy for a whole team.

Another challenge is the visual analysis of the environment of the robot, in order to detect key features like the bal and the opponents. This analysis is performed using small neural networks.