Design of Multi-Scroll Memristive Hopfield Neural Network and Application to Synchronous Coverage Search

This paper proposes a multi-scroll memristive Hopfield neural network (M-HNN) and applies it to airground collaborative synchronization search tasks. To be specific, a hyperbolic-tangent memristive synapse is first constructed to generate a chaotic multi-scroll M-HNN with a controllable number of scroll attractors, and the memristor flux is then employed as the driving signal for mobile robot motion control. By exploiting the ergodic property of the designed multi-scroll M-HNN, the search capability of the chaotic mobile robot is enhanced. When the air-ground collaborative search task is further considered, the problem can be reformulated as a synchronization tracking control problem for a chaotic mobile robot guided by a chaotic unmanned aerial vehicle (UAV). To address this problem, a critic-only integral reinforcement learning (IRL)-based synchronization strategy is developed to minimize the synchronization tracking error. Finally, the dynamics of the multi-scroll M-HNN are analyzed, and the effectiveness and superiority of the proposed synchronous coverage search algorithm are validated.