Research On Motion Control Approaches For An Intelligent Mobility Aid

With the continuing development of China's population aging, elderly market makes growing demands and increasingly high requirements on walking aids. Conventional canes and walkers, which are tiring and not safe enough for the elderly, cannot fully satisfy the demand of safety and comfort any more. It will be of great interest to introduce robotic technology into a conventional walking support system.A prototype of an intelligent mobility aid is developed in this paper to help the elderly walk easily, safely and independently in the outdoor environment, including the design of differential driven mobile system, human-walker interaction system, environment sensing system and motion control system. Two different motion control approaches are proposed.The first one is an intent-based passive control approach. It employs nearest neighbor algorithm to extract stable walking intent from user's fluctuating operations, activates a smooth control by an intent-based user-oriented interaction model, and fuses the environmental information in a passive way.The second one is an online learning control approach based on CMAC neural network. It forms a closed-loop feedback control by interpreting the operations on the handle bars as transformed error signals, employs time inversion technique to acquire the online teacher signal, and learns user's operating habits in a boot-strapping manner.Experimental results show that the intent-based passive control approach can improve the comfort during the different phases of motion and prevent the mobility aid from colliding obstacles or falling into stairs; the CMAC-based online learning control approach can make the user-walker interaction friendly to some degrees.