MPC technique is a tool of automatic Science to compute an optimal control. However, it does not handle long horizon nor uncertainty. The idea here is to develop an MPC-control ros node that could be supervised.
Potential Supervision: Guillaume Lozenguez
Multi-robot path planning
The difficulty in Multi-robot path planning is to find several paths that permits the robot to reach their goal position by avoiding any collision.
Q-Learning is a simple but powerful tool permitting an artificial agent to learn autonomously how to act. The goal of this project is to apply this algorithm to mobile robots.
