A Real-Time Trajectory Generator With 3rd-Order Constraints Satisfaction Through Velocity Mode Evolution

This article proposes a new real-time trajectory generator, RTG-VM, aimed at enabling a robot to respond quickly and smoothly to sudden external commands and adjust the original trajectory in real-time. It is based on velocity mode evolution, which rapidly plans time-optimal trajectories between arbitrary initial and final motion states. Initially, two velocity mode (VM) planning methods without considering the target displacement are studied: time-optimal VM and reverse initial jerk VM. Subsequently, the VM acceleration profile is adjusted with minimal time cost to reach the desired target displacement. Finally, an effective topological structure of the evolution of acceleration types over planning times is presented. In the experiments, the computation time of RTG-VM for a single axis is about one microsecond. Its real-time performance is at least 38% better than that of Ruckig. Finally, an experiment is conducted on a 5-degree-of -freedom hybrid robot passing through the vicinity of a singular configuration.