Data-driven double-level motion planning for redundant robots
Redundant robots play an important role in intelligent manufacturing,rescue and relief work,and space exploration.However,motion planning of redundant robots with unknown structural information is a tricky problem.In addition,if the orientation of the end-effector of a redundant robot is not considered,a task may fail due to the change of orientation.To solve this problem,a data-driven double-level motion planning scheme is proposed,which is formulated at the joint acceleration level to minimize the norm of the joint velocity,maintaining the orientation of the end-effector and considering the multiple physical constraints of joints.Furthermore,a discrete neural dynamics solver is constructed to solve the proposed scheme online.Theoretical analysis,computer simulations,and experiments verify the feasibility and effectiveness of the proposed data-driven double-level motion planning scheme for redundant robots.
NETL
NSTL
万方数据
