Adaptive asymptotic switched tracking control for a cross-media vehicle

This paper develops a reinforcement learning-based adaptive asymptotic switched trajectory tracking control strategy for a cross-media vehicle (CMV) suffering from unknown hydrodynamics and external disturbances. A novel switched system framework is reported for the modeling of the vehicle. To better handle the unknown hydrodynamics and optimize the control performance, a reinforcement learning (RL) methodology is presented. The adaptive technique is then introduced to estimate the external disturbances. Subsequently, the asymptotic switched tracking control scheme is designed for the CMV. Compared with existing approaches, the proposed method has the following merits: it develops the switched system framework and switched control synthesis for CMV systems, thereby mitigating modeling and control conservatism; the RL strategy offers the control algorithm with effective compensation for unknown hydrodynamics, while giving optimized control performance; moreover, the asymptotic tracking performance rather than the bounded tracking responses is obtained. Lastly, the simulation is run to demonstrate the scheme’s effectiveness and superiority.