A control method and system for a dual-arm folding clothes robot

Abstract

This invention provides a control method and system for a dual-arm folding robot, belonging to the field of robot control technology. The control method constructs a perception-to-control closed loop by fusing visual, force, and tactile multimodal data. Tactile and force sensing compensate for the blind spots of vision in self-occlusion scenarios, improving the perception of grasping, sliding, and folding. A joint representation of clothing based on a visual semantic network structurally encodes geometric shapes, semantic components, and confidence levels, providing a robust state for planning that integrates prior knowledge and real-time observations, reducing decision-making errors caused by deformation and noise. A hierarchical strategy decomposes long-term folding into sub-objective sequences and uses multimodal data to correct the action closed loop, achieving coordinated control of grasping, folding, and folding. Finally, torque commands are generated based on the control algorithm to ensure smooth and safe interaction, comprehensively improving the folding success rate and operational stability of the dual-arm folding robot under different materials, sizes, and initial shapes.

Images