Research On Adaptive Gait Planning Method For Paraplegic Walking Exoskeleton In Multi-terrain

The research on control of lower-limb walking-assist exoskeleton robots is mainly aimed at fixed rehabilitation training scenarios such as flat ground,slopes and stairs.In these scenarios,the research on the gait planning method of the walking exoskeleton is mainly oriented to the scene environment with known terrain parameters,which has been relatively mature.However,it still remains unsolved that how we can plan the gait of the lower limb walking aid exoskeleton in variable-parameter terrain scenes and complex mixed terrain scenes.Therefore,thesis proposes a lower extremity exoskeleton gait planning method,which is suitable for multi-terrain environments and makes the system better adapt to human-machine-environment,so that it not only fits laboratories and rehabilitation hospitals,but also scenes in our real life.The main work of thesis can be summarized as follows:1.In order to solve the terrain parameter adaptation problems of the lower limb walking-assist exoskeleton robot in complex terrain scenes,a gait end trajectory planning model has been proposed.It is based on the improved Kernelized Movement Primitive(KMP)and can handle some current problems.For example,the gait trajectory planning based on the dynamic motion primitive model can hardly establish a unified gait planning framework,since it's over-reliant on a single teaching and lacks process point constraints.What's more,in thesis,an adaptive generation algorithm of task constraint points suitable for multi-terrain conditions,like stairs and slopes,has been proposed.It is based on the establishment of a prior probability distribution reference for the teaching curve of the high-dimensional gait end trajectory and is expected to realize the multi-terrain adaptive swing phase end trajectory planning.Besides,combined with the linear inverted pendulum model and the time series estimation of gravity center trajectory of exoskeleton,an inverse kinematics model of lower limb exoskeleton has been established,the joint space gait trajectory has been figured out and the gait planning problem of the lower limb walking exoskeleton in a single complex scene has been solved.2.On the basis of improving the position-space gait trajectory planning model of nuclearized motion primitives,a gait planning method suitable for the tasks in transitional terrain scenarios has been proposed,in order to solve the problem that the paraplegic walking-assisted exoskeleton robot relies on the task mode of the sub-state machine and lacks gait planning in terrain transition scenarios.Furthermore,based on the parametric modeling representation of tasks in typical transitional terrain,the planning stages in transitional terrain scenarios have been divided,and the gait end trajectory planning in preparatory and transitional stages as well as the adaptive estimation of critical points of stages has been discussed,through which the gait planning of the tasks in transitional terrain scenarios has come true.3.The paraplegic walking-assist exoskeleton robot system platform as a foundation of thesis has been introduced.On the basis of the prototype system,a depth vision module for processing terrain parameter information has been added.The hardware and software framework of the system has been figured out and experimental verification scenes,including obstacle terrain and flat ground-stair transitional terrain,have been built.It has been verified that the methods put forward in thesis are effective given the comparison between the natural gait value collected by the motion capture system and the planning results of the dynamic primitive model.