Comparison of self-moving cart motion control methods for collision avoidance problems

This paper discusses the self-moving cart collision avoidance problems as an example of the rigid body motion control problems which have time-varying ideal states. This paper mainly discusses the theoretical issue by numerical simulations and one-dimensional self-moving cart model is established for this purpose. The authors apply 1) MPC (Model Predictive Control) using C/GMRES (Continuation/Generalized Minimal Residual) algorithm, 2) UFSC (Updating Final-State Control), and 3) PID (Proportional-Integral-Derivative) control with safe brake constraints to this problem. The characteristics of each method are discussed through the comparison of the simulation results. For this purpose, the authors set three different conditions. In case 1, the controller knows the time-varied motion of the obstacle exactly. In case 2, the obstacle's movement varies suddenly during the simulation. In case 3, the collision avoidance to the stationary obstacle is considered. The controller knows the exact obstacle's motion beforehand. All methods achieve the collision avoidance successfully in all simulation conditions. The comparison result shows some advantages of the UFSC in the computation time, the acceptable control performance, the efforts of the control parameters tuning, and the performance retention for the prior information.