Fast approach for robot motion planning

This paper describes a new approach to robot motion planning that combines the endpoint motion planning with joint trajectory planning for collision avoidance of the links. Local and global methods are proposed for end-point motion planning. The joint trajectory planning is achieved through a pseudoinverse kinematic formulation of the problem. This approach enables collision avoidance of the links by a fast null-space vector computation. The power of the proposed planner derives from: its speed; the good properties of the potential function for end-point motion planning: and from the simultaneous avoidance of the links collision, kinematic singularities, and local minima of the potential function. The planner is not defined over computationally expensive configuration space and can be applied for real-time applications. The planner shows to be faster than many previous planners and can be applied to robots with many degrees of freedom. The effectiveness of the proposed local and global planning methods as well as the general robot motion planning approach have been experimented using the computer-simulated robots. Some of the simulation results are included in this paper.