Optimization-based Model Predictive Tube Control for Autonomous Ground Vehicles with Minimal Tuning Parameters

Abstract

The trajectory planning and tracking problem are critical points of intelligent vehicles concerning their safety and stability. If these parts are separated, interactions should be made between them, especially when sudden changes and disturbances appear. This paper presents a method that integrates the two parts using a cascade structure. Additionally, the proposed method deals with the interaction between the lateral and the longitudinal trajectory based on dynamical considerations. The whole problem is handled using the model predictive method based on online optimization. This system receives the path borders as input and generates the control requests for the actuators on its output. The configuration space of the system can be maximized to gain stability by handling the lateral-longitudinal parts and the trajectory planning and tracking in one complex system. The main advantage of the proposed approach is that the optimization problem in the predictive method is formulated so that the path and dynamics are considered equality and inequality constraints, and the cost function includes only a physical phenomenon to be minimized without tuning parameters. The evaluation of the proposed algorithm is presented in this paper based on simulation and real-time measurements results.