This is the complementary website for the paper. In the paper, we address the challenge of cooperative navigation for multiple robots in a crowded environment by a control framework combining game-theoretic model predictive control and Social-LSTM human trajectory prediction model. We employ an iterative best-response approach to develop two algorithms for solving the MPC problem in centralized and distributed manners. We validate the effectiveness of the control framework by simulations using CrowdNav environment.

An example of multi-robot cooperative navigation among human pedestrians can be illustrated in the below picture. The robots need to reach their goals and avoid collision with other robots and human pedestrians. Moreover, since the robots move in the same direction,  they can coordinate for moving in a flock while navigating to the goals. Our hypothesis for this scenario is that, if the robots move in a flock, the disruption and discomfort to the human pedestrians caused by the robots' motion can be mitigated.