Model Predictive Control-Based Lateral Control of Autonomous Large-Size Bus on Road with Large Curvature 2021-01-0099

This paper describes a lateral control of autonomous large size buses on road with large curvature. In the case of long and wide commercial vehicle such as large bus, applying centerline tracking controllers in constrained environments such as large curved road (e.g. turning at intersection) may cause some concerns. Two concerns are considered: inner lane crossing related to collisions with curb and opposite lane crossing related to threatening surrounding vehicles. Considering relations between width and curvature of the road and length and width of the large size bus, the curvature of road at which inner or outer lane crossing begin to occur was calculated when centerline tracking controller was applied. Thus, the proposed algorithm optimizes motion of the bus by using model predictive control (MPC) using road geometry as constraints. Based on geometric relations of curved road and vehicle, distance from the lane to each corner of the vehicle is defined using relative lateral position and relative heading angle of the vehicle and road center line, which is used in the MPC formulation. A slack variables are used to solve feasibility problem caused by the difference between open loop prediction and closed loop trajectory in receding horizon optimal control. Performance indexes are defined to evaluate performance of the algorithms. The proposed algorithm was evaluated via computer simulation. The performance of the proposed algorithm was compared with the centerline tracking controller. It is shown that the proposed algorithm allows the large size bus to cope well with steering on road with large curvature.