Game-model prediction hybrid path planning algorithm for multiple mobile robots in pursuit evasion game

In this paper, a novel hybrid path planning algorithm is proposed based on the game theory and model predictive control method. Firstly, we use the feasible predicted positions, which are calculated by inputting discrete steering angles to the evader model, as the game decision set of the evader. Then, the pursuer's game decision set is obtained based on the model predictive control method, which takes each predicted position as a virtual target and solves the suboptimal problem with input constraints. Finally, we design the payment function, and acquire the collaborative strategy by solved the pursuit-evasion problem based on the zero-sum game decision method. Numerical simulations show that the proposed method can capture the evader with different maneuvers effectively. From further simulation analysis, it can also be seen that, through the proper allocation between the item of capture distance and the item of team distribution in the payment function, the evader can be captured faster while maintaining the team capture situation.