Hybrid Model Predictive Control for Unmanned Ground Vehicles.

This paper presents a hybrid approach to enhance the path planning for unmanned ground vehicles (UGVs) by combining stochastic dynamic programming (SDP) based hybrid model predictive control (HMPC) with Dijkstra-based pseudo priority queues (PPQ). The proposed approach employs a platform-specific model that takes into account skid-slip effects, a global cost-to-go (CTG) function, and dynamic obstacles. To achieve efficient path planning on large maps, the proposed approach employs a Dijkstra algorithm and PPQ to generate the CTG function. Moreover, the HMPC process incorporates the latest CTG information, the vehicle model, and the perceived environment, encompassing both static and dynamic obstacles. Extensive simulations have been conducted to comprehensively evaluate the performance of the proposed technique. Through a comparative analysis against existing path planners, the results demonstrate the effectiveness and superiority of the hybrid approach.