Design of Robot Motion Control Strategy for Multi-Stage Tasks

Robots, as essential tools for advancing modern society, not only drive the growth of the manufacturing industry but also enhance the quality of life. However, with the increasing demands of society and the need to perform long-term, complex tasks, traditional structured control methods limit their applicability to multi-stage tasks. To address this challenge, this paper proposes a generalizable robot motion control strategy based on the stochastic configuration networks, with a focus on the generalization of control strategy parameters, thereby enhancing the strategy’s applicability for non-professionals. Specifically, the generalization problem of multi-stage robotic trajectory tracking is addressed via learning from demonstrations. The stability of the control strategy is analyzed, and the corresponding parameter constraints are derived. Importantly, the proposed strategy does not require parameter readjustment when facing new task trajectories. Finally, the generalization ability of the proposed strategy is validated through real scenarios.