Safety-aware isoline tracking control for unmanned surface vehicle

Abstract

The nmanned surface vehicle (USV) are increasingly employed in complex marine environments for various missions. However, challenges such as weak GPS signals and multi-obstacle scenarios can degrade control performance and introduce significant safety risks, potentially leading to mission failure. In this backdrop, this paper proposes a safety-aware isoline tracking control scheme for USV. Consequently, this scheme eliminates the position control layer by introducing a distance-based sensor strength, achieving precise tracking through yaw adjustments. For the safety concern, an input-to-state safe control barrier function (CBF) is incorporated, ensuring reliable obstacle avoidance. Additionally, the impact of lumped uncertainties, including model inaccuracies and external disturbances, is mitigated using a reinforcement learning-based echo state network (ESN). This network is trained to effectively compensate for these uncertainties in real-time manner. Finally, theoretical analysis and simulations are conducted to verify the effectiveness of the proposed scheme.