Adaptive tracking control and torque allocation for multi-wheel independent drive vehicle in off-road terrain based on MAS

This paper proposes a non-hierarchical adaptive trajectory tracking control and torque allocation method based on a multi-agent system (MAS) framework for multi-wheel independent drive vehicles (MWIDEVs) in off-road scenarios. To avoid high-dimensional optimization issues caused by multiple actuators in MWIDEVs, we first transform the commonly used centralized or hierarchical tracking and torque allocation problem into a distributed multi-agent control problem. Both the vehicle dynamics model and wheel dynamics models are treated as individual agents, achieving coordination through inter-agent communication. Then, a distributed wheel-terrain interaction estimation method is developed for each agent to estimate its maximum friction limit at the wheel-terrain interface. To achieve this, in vehicle dynamics agent, the nominal vertical load is calculated by neglecting terrain-induced disturbances. Based on the nominal vertical load, we design a fixed-time observer and a load redistribution method for each wheel to estimate vertical load variations caused by off-road terrain at tire-terrain contact patches. The estimated tire force limits are transmitted to the vehicle agent to generate reasonable desired control torques, thereby avoiding situations in which the actual torques exceed the ground-adhesion limits. Finally, we established two off-road terrains in TruckSim with randomized potholes, corresponding to Class C and E road roughness according to ISO 8608, and validated the algorithm's effectiveness and feasibility through MATLAB-TruckSim co-simulation as well as hardware-in-the-loop (HIL) tests.