Dual heuristic programming with just‐in‐time modeling for self‐learning fault‐tolerant control of mobile robots

Abstract The development of autonomous wheeled mobile robots (WMRs) requires designing high‐performance motion controllers under complex dynamical constraints, which has been an important research topic in the fields of control theory and robotics. Considering the faults or disturbances in practice, previous works have designed various stable fault‐tolerant tracking control methods for WMRs. However, it is still challenging to design an optimal or nearly optimal controller in the case of faults. In this article, in order to deal with the changed properties of WMR dynamics, a just‐in‐time learning (JITL) based dual heuristic programming (DHP) algorithm is proposed to optimize the control performance under faults or disturbances. Without knowing the information of faults and the dynamics model, the changing dynamics can be identified online using the data‐driven method, and the control performance is optimized via DHP, which is a class of online reinforcement learning approach. The effectiveness of the proposed method is verified through fault‐tolerant tracking control simulations of a WMR system under different fault modes. It is shown that the proposed JITL‐based DHP approach has much better performance than previous DHP methods without fault estimation.