Human-Following Control of Cane-Type Walking-Aid Robot Within Fixed Relative Posture

Assisting and supervising daily long-durational walking is very crucial for patients with lower extremity dysfunction, especially in the stage of recovery towards a state of walking independently. However, due to the shortage of caregivers and high-cost of nursing, long-term manual assistance and supervision is costly. Thus, in this paper, we propose a cane-type walking-aid robot to follow a human user for the safety and supervision of independent walking during rehabilitation training. Strongly motivated by clinical suggestions and user experiences, the human-following rule that the robot should follow the user in the required position within a fixed relative posture (keeping a certain distance in front of the user, 15-20 cm lateral to the healthy side, and 0 to the user's orientation) is established. To guarantee a reliable following, a quantitative human walking intention estimation method is proposed using the on-board Laser Ranger Finder (LRF) and Kalman Filter (KF). To effectively monitor the user's walking, a finite-time control method for the cane robot is also proposed to ensure its human-following performance. Experiments were conducted to demonstrate the effectiveness of the proposed system and methods. The experimental results show that the proposed system can obtain satisfactory performance of human-following and can be made available for different users in different walking modes.