Robot Joint Angle Mapping for Accurate Human Pose Imitation

The patent segments the motion imitation task into distinct components: joint detection (identifying human joint positions), angle calculation (computing joint angles from detected positions), and mapping learning (creating correspondence between human and robot joint angles). This segmentation allows the robot to systematically acquire motion imitation capability without compromising its autonomous behavior capabilities.

The patent introduces an intermediary mapping function that translates human joint angles into robot joint angles. This mapping acts as a mediator between human motion input and robot motion output, enabling the robot to learn and reproduce human motions while maintaining its independent autonomous decision-making capabilities.

The patent replaces complex mechanical motion transfer mechanisms with a software-based mapping learning system. Instead of using physical mechanisms to transfer motion from human to robot, the system uses computational algorithms to detect joint positions, calculate angles, and learn the mapping relationship, thereby achieving accurate pose imitation with reduced mechanical complexity.

The patent transforms the pose imitation problem from a spatial coordinate transformation problem into a parameter-based angle mapping problem. By focusing on joint angles as the key parameters and learning the mapping between human and robot joint angles, the system achieves accurate pose imitation while simplifying the computational burden compared to full 3D coordinate transformation.