A Receding Horizon Online Trajectory Generation Method for Time-Efficient Path Tracking with Dynamic Factors

Current path-tracking trajectory planning methods intrinsically suffer from a heavy computational burden, hindering them from modern autonomous robotic applications requiring real-time reactivity. To achieve flexible, accurate, and efficient motions, this article proposes a real-time trajectory planning framework for time-efficient path tracking. First, a receding horizon method is designed that generates local trajectories online while considering the global kinematic constraints. Second, an analytical closed-form method is developed for calculating the local time-minimal trajectory. Third, the models and solutions are established for handling dynamic factors. Compared to existing methods, this method exhibits lower computational complexity and can deal with path changes during motion executions while maintaining tracking accuracy and time efficiency. Experiments using Franka-Emika Panda robots are conducted in handover scenarios involving unpredictable dynamic obstacles and human interventions. The results demonstrate the low computational overhead. The robot flexibly reacts to online path changes, maintaining tracking accuracy while marginally compromising time efficiency.