A Pseudospectral optimal motion planner for autonomous unmanned vehicles

This paper presents a pseudospectral (PS) optimal control algorithm for the autonomous motion planning of a fleet of unmanned ground vehicles (UGVs). The UGVs must traverse an obstacle-cluttered environment while maintaining robustness against possible collisions. The generality of the algorithm comes from a binary logic that modifies the cost function for various motion planning modes. Typical scenarios including path following and multi-vehicle pursuit are demonstrated. The proposed framework enables the availability of real-time information to be exploited by real-time reformulation of the optimal control problem combined with real-time computation. This allows the each vehicle to accommodate potential changes in the mission/environment and uncertain conditions. Experimental results are presented to substantiate the utility of the approach on a typical planning scenario.