The Research Of Motion Control And Safety Custody Strategy For Walking-aid Robot

The mobility of elderly degrades with age, which affects not only their daily life, but also the life quality and causes dependence of other in their daily life. Now China gradually goes into the aging society. It is estimated that the population of elderly (over 60) will reach to 33% by 2050s. Elderly care work requires a great of medical care, incurs large costs. So it is extremely important to develop the devices, which can support and assist elderly in their daily life. The walking-aid robot which applies robotic technologies can help elderly to restore the ability to walk, get the chance of independent and improve the quality of their life, which are very important to the rehabilitation care system of forthcoming elderly society. Here we investigate the motion control and safety custody strategy for walking-aid robot. Aiming at the requirements of safety and compliance for walking-aid robot, reinforcement learning-based shared control algorithm and multi-sensor fusion based compliance motion control are proposed to improve the compliance of walking-aid robot. Two fall detection methods are proposed, and user fall prevetion motion control is realized to raise the safety of walking-aid robot.The development and design process of intelligent walking-aid robot I and II are introduced in detail. Firstly, aiming to the requirements of considering both the user's operating feeling and robot motion control, intelligent walking-aid robot I is developed. On the basic of intelligent walking-aid robot I, adjustable fence based auxiliary support structure is added to realize effective physical support, improve human-robot interface system, and a human-robot interaction force control based admittance control for walking-aid robot is proposed to develop intelligent walking-aid robot II, which is safer, more stable and flexible with lower cost, and easy to realize productization.A reinforcement learning-based shared control algorithm for intelligent walking-aid robot is proposed. Considering user control efficiencies and walking environments, reinforcement learning method is adopted to adjust user control weight on line. Shared control method is applied to synthesize user motion intention and planned velocity of robot, get the motion control which not only compliances user's intention, but also keeps user's safety. The experiments verify that the reinforcement learning-based shared control algorithm can ensure user movement performance and smoothness with strict safety requirements.The user fall detection and prevention strategy for intelligent walking-aid robot is studied. Wearable sensors are adopted to observe the user walking state. The distance between the human Center of Pressure (COP) and the midpoint of the human feet and the height of user's waist are assumed to be significant features to detect the fall events. During normal walking, the robot is controlled by a conventional admittance control scheme. When the tendency of fall is detected, robot will immediately react to prevent the user from falling down. Experiments demonstrate validity of the proposed user fall detection and prevention strategy.Aiming to compliance and safety in the motion control of intelligent walking-aid robot, a multi-sensor fusion based walking-aid robot compliance motion control and safety custody method is proposed. Force sensors and laser range finder are used to get the user motion intentions of upper limbs and lower limbs respectively, then Kalman Filter based Sequential Probability Ratio Test(SPRT) method and decision function were used to detect falling and falling mode. When user is normal walking state, the two user motion intentions are fused by multi-sensor fusion technology to get more accurate and compliance user motion intention. When the tendency of a fall is detected, robot brake imminently. Experiments pove that the proposed algorithm can improve the compliance motion control, detect fall tendency and prevent user from falling effectively.The user fall prevetion motion control of intelligent walking-aid robot is studied. To improve safety and compliance in the process of fall prevetion, an impedance-sliding mode control based user fall prevetion motion control algorithm is proposed. Firstly the walking-aid robot dynamics model and impedance control model while user is falling are established. Then combining sliding mode control algorithm, design a dynamic compensator to realize impedance control when robot system reaches to sliding mode surface. The MATLAB simulation experiments and robot experiments prove validity of the user fall prevetion motion control algorithm.