An Iterative Learning-based Integrated Motion Planning and Control Method for Autonomous Patrolling of Unmanned Surface Vehicles
Yun Tang,Fei Wu,xinle Gong,Chao Chen,Hongliang Liu,Jie Ma,Yi Qin,Huayan Pu,Jun Luo
DOI: https://doi.org/10.1088/1361-6501/ad7b66
IF: 2.398
2024-09-21
Measurement Science and Technology
Abstract:The minimization of maritime patrol durations and the optimization of patrol trajectories are still a prevailing challenge for autonomous navigation of Unmanned Surface Vehicles (USVs). In this paper, we propose an integrated motion planning and control method to achieve time-optimal autonomous patrol of USVs by a data-efficient iterative learning-based predictive control (ILPC) algorithm. We build the optimization problem of the algorithm by introducing a local cost and a local safe constraint set for closed-loop efficient data-driven learning. To guarantee recursive feasibility, stability and convergence properties of the optimization algorithm at each iteration, the local cost and the local constraint set are designed and update iteratively using the historical vehicle states at previous iterations as a dataset. Different from traditional optimization methods, the proposed method does not require a reference path. The cost function of the optimization method decreases monotonically and converge to obtain a time-optimal control law for trajectory planning and tracking of USVs only after several iterations. The proposed approach is validated on the robotic operation system (ROS) in typical maritime patrol scenarios. The results show that the proposed method implement a 5% reduction of patrol time, fewer cross-track errors, and smoother trajectories compared to the traditional MPC algorithm under water flow conditions. In addition, Enhancements in patrol operation efficiency of USVs are achieved by the algorithm, even under the constraints of varied patrol paths, attesting to its versatility.
engineering, multidisciplinary,instruments & instrumentation