Whole-body Control Of Humanoid Wheeled Robot Based On Hierarchical Neural Dynamics And Continuous Task Transition

Humanoid wheeled robot has a good development prospect in various fields such as household service and industrial production due to its strong operation ability and mobility.However,due to its strong coupling and redundancy,traditional methods are difficult to make reasonable planning when completing complex tasks.Therefore,this dissertation takes task as the basic unit of robot execution and studies the multi-task control of robot,so that humanoid wheeled robot can plan tasks with different priorities.The research content of this dissertation can be summarized as follows:(1)A trajectory tracking control scheme of humanoid wheeled mobile robot based on adaptive neural network control is designed.By using neural network to approximate the uncertain terms in the model,the problem of poor control effect caused by the inaccurate parameters of the dynamic model can be solved,and the stability of the control system is proved.(2)The whole body control framework based on hierarchical quadratic programming method is established to realize the priority multi-task control of humanoid wheeled robot.This method can solve the task stack with defined priority,and can handle the task with equality and inequality constraints.(3)An adaptive quadratic programming solver based on neural dynamics is proposed to solve hierarchical quadratic programming problems quickly.By introducing compensation and adding adaptive control term,this solution method can deal with equality and inequality constraints,and has stronger robustness and accuracy.The neural dynamics solver is applied to solve the whole body control problem based on hierarchical quadratic programming on-line,and then controlled by adaptive neural network controller,and the experimental verification is given.(4)A scheme based on transition variables is used to deal with the discontinuous output speed of the whole-body control framework during task insertion,removal and task priority swapping of multi priority tasks in the whole-body controller.This method can deal with not only equality constraints,but also inequality constraints,and it has good transition effect without changing the control structure.