Abstract: Human transfer task is one of the labor-intensive nursing tasks, so robotization of transfer nursing tasks is an urgent need in the nursing field, and dual-arm robots are gaining attention in the nursing field for their flexibility and versatility. Unlike traditional industrial robots for locating tasks, nursing tasks require physical human-robot interaction, and the scenario is usually an unstructured environment with obstacles, so human safety and comfort during the task are critical. To realize the transfer-care task of the dual-armed robot in an unstructured environment, a Trajectory Deformation Algorithm (TDA) that equip the robot with the functions of obstacle avoidance and active compliance is proposed. Firstly, the repulsive and attractive functions are designed based on the concept of artificial potential field, then the obstacle avoidance and active compliance function be described by the forces of the field and the human-robot interaction forces. Thus, the equivalent motion of the virtual spring-damped system under the forces is considered as the trajectory deviation, and the trajectory deviation is transformed to the joint space by the Jacobi matrix to avoid the inverse kinematic solution of the redundant robot. Simulation results show that the TDA can achieve active compliance and effectively avoid obstacles. Finally, the transfer-care task of dual-arm robot is conducted, which verified that the TDA can adjust the trajectory in real-time based on interaction forces and obstacle information. The achievements are of great significance to prove the TDA is suitable for nursing-care tasks in unstructured scenes, improve the security during the human-robot interaction, and realize the promotion and application of the robot in the field of rehabilitation nursing.

Key words: transfer-care, dual-arm robot, trajectory deformation, active compliance, artificial potential field