Quasi-optimal energy path planning for anthropomorphic manipulator using gravity torque mapping

In this paper, the energy used to hold the position of a robot arm against gravitational torque is presented. As the gravitational torque is static over the pose of the manipulator, it is possible to draw an energy map in a 2D grid. To reduce the energy used in a point-to-point movement, a modified Dijkstra's path-finding algorithm is used. Generally, Dijkstra's algorithm finds the shortest path on the grid, but in this paper, it is modified to generate the path with minimum use of position-holding energy. After generating the joint path off line, the energy use is compared between a conventional path and the proposed method is evaluated in a simulation including full dynamics.