Abstract

Stewart parallel mechanisms are widely used in motion simulation, posture adjustment, and active compensation owing to their high stiffness, high load capacity, and high positioning accuracy. However, under unknown paths or time-varying target trajectories, their strong multi-chain coupling, limited workspace, and configuration-dependent motion capability may cause actuator velocity and acceleration violations, especially near workspace boundaries or low-mobility regions. Moreover, conventional online trajectory scaling methods are usually designed for acceleration bounds with regular positive and negative limits. After nonlinear kinematic constraint mapping, the acceleration bounds of a Stewart mechanism in the path-parameter space may become same-signed, shifted, or abruptly varying, which can lead to planning failure or motion discontinuity. To address these problems, this paper proposes a condition-number-aware velocity bound regulation scaling method for online trajectory planning under unknown paths. Path–velocity decomposition is first used to transform the six-degree-of-freedom trajectory planning problem into a path-time-law planning problem, and actuator constraints are mapped into the path-parameter space. Then, a workspace classification model and the dimensionless Jacobian condition number are introduced to evaluate local motion capability. Based on the workspace level and condition number, a hierarchical scaling factor is designed to adaptively adjust the path velocity boundary. In addition, the conventional trajectory scaling method is improved by introducing a special acceleration-bound handling mechanism. Single-axis and platform experimental results show that the proposed method can effectively handle special acceleration bounds, reduce actuator constraint violations, and improve the safety and continuity of online trajectory planning.

Keywords