Time-varying Multi-robotic Fish Formation Tracking Using Distributed Nonlinear Model Predictive Control

Fish schooling behavior has garnered significant research interest in recent years due to its biological relevance and potential engineering applications. In this paper, we propose a novel distributed nonlinear model predictive control (NMPC) framework to address the time-varying formation tracking control problem in tail-actuated multi-robotic fish systems. First, a robotic fish control model is developed by integrating a dynamic model with a central pattern generator (CPG) model, where the caudal fin oscillations directly serve as control inputs. Second, dynamic reference trajectories for each robotic fish are generated based on consensus errors and the gait features of both neighboring. To enhance tracking performance, a line-of-sight (LOS) strategy is incorporated into the NMPC framework. Finally, the proposed method is validated through numerical simulations. The results demonstrate the method's effectiveness, robustness, and scalability, offering a promising solution for distributed control in multi-robotic fish systems.