Automatic Steering Control of an Unmanned Transport Truck in Off-road Environments

Abstract

In structured environments, steering control of autonomous driving is almost performed by using localization based on lane markers. However, in off-road environments, there are no lane markers and the road boundary is detected by some sensors. The boundary is not smooth and automatic steering control based on the detected boundaries has problems. In this paper, path following control with the nonlinear weight of lateral errors is proposed. The lateral error is derived from the boundary which is detected by using deep learning. In conclusion, it was confirmed by actual vehicle experiments, the proposed method worked well on agricultural roads.