abstract
- Model Predictive Control (MPC) is a powerful tool for robotics. However, its high computational load poses a challenge for embedded systems. This paper proposes an efficient MPC with a Proximal Gradient method approach (PG-MPC) for controlling a Mecanum wheel robot in an obstacle-constrained environment. Comparing the proposed PG-MPC to a baseline MPC, two case studies are considered for simulation results. First, four representative points were set as goals in an obstacle-free environment. For the second case study, two obstacles are imposed along the trajectory leading to the goal of the vehicle. It is observed that the trajectories generated by both controllers are similar. However, under the PG-MPC architecture, the computational time is significantly improved compared to the MPC baseline.