In order to operate in a clustered environment, a rehabilitation robot should have many degrees of freedom (DOF) and it should be redundant. Having a 7 DOF manipulator, the robotic system can skillfully support the user in every day life, but this introduces the problem of planning robot motions in real time. This paper introduces an efficient motion planning method for the 7 DOF manipulator. In order to avoid the computation complexity, typical for multidimensional joint space, the planning takes place in the Cartesian space. By using the graph based search, in combination with the fast inverse kinematics and with the checking of the manipulators ability to move without collision in each step, the approach is capable to escape the workspace dead-ends and it is also efficient enough for the practical usage. The resulting path is then smoothed or passed to the local planning method. Furthermore, the proposed approach enables neglecting the distant obstacles in the calculation of distances used in collision checking, which additionally decreases the overall computation time. The approach is implemented on the 7 DOF manipulator within the rehabilitation robotic system FRIEND II.
展开▼
