A Time-Optimal Trajectory Planning Strategy for an Aircraft With a Suspended Payload via Optimization and Learning Approaches
Hean Hua, Yongchun Fang, Xuetao Zhang, Qian Chen
Abstract
In this article, a real-time interactive time-optimal trajectory planning (TOTP) strategy is proposed for aircraft slung-load systems, in which the practical performance is guaranteed by taking advantage of convex optimization and reinforcement learning (RL) techniques. To the best of our knowledge, it is the first TOTP solution for an aircraft suspension system subject to path, velocity, acceleration, and cable tension constraints, where a practice-oriented RL policy is designed to interact with the system online. Specifically, by exploring the differential flatness property of the system, the states are projected into path coordinate space, where all the physical constraints are formulated in the TOTP with convex forms. Then, the TOTP is transformed into a large sparse optimization problem after discretization, which can be efficiently solved using convex solvers. Subsequently, a deep RL network is designed to optimize the TOTP results, where efficient learning is achieved in the solution space, thereby guaranteeing practical reliability and strong robustness via online interaction in applications. In addition, by incorporating convex optimization and learning techniques, the gap between simulation and experiment is successfully bridged, where the physical feasibility and efficient learning are ensured in the framework. Finally, comparative simulations and experimental results are included to show the effectiveness of the proposed strategy.