Cooperative optimal path planning for herding problems

Abstract

In this thesis we study a new type of pursuit-evasion game, which we call the herding problem. Unlike typical pursuit evasion games where the pursuer aims to catch or intercept the evader, the goal of the pursuer in this game is to drive the evader to a certain location or region in the x-y plane. This herding model is proposed and represented using dynamic equations. The model is implemented in an effort to understand how two pursuers work cooperatively to drive multiple evaders to the desired destination following weighted time-optimal and effort-optimal control paths. Simulation of this herding problem is accomplished through dynamic programming by utilizing the SNOPT software in the MATLAB environment. The numerical solution gives us the optimal path for all agents and the corresponding controls as well as the relative distance and angle variables. The results show that the pursuers can work cooperatively to drive multiple evaders to the goal.