Robot motion planning using potential fields
Par Dunias · TU/e Research Portal · 1996
The problem addressed in this paper is the autonomous motion planning of robots in known environments and with unknown, possible moving, obstacles. The motion planning is of the point-to-point type where the main objective is to reach a given goal pose of the robot without colliding to any obstacles. Robot Motion Planning is considered, aiming autonomy for articulated robots in real-time applications. The motion planning method employed is using Artificial Potential Fields. The artificial potential field is a function whose gradient is used to calculate a torque applied to the robot, enforcing it to reach its goal while avoiding obstacles. Apparently this potential field depends on the goal state of the motion planning and the geometry of the obstacles around the robot arm. The artificial potential field has to attain its maximum values at the obstacles and its global minimum at the goal state. The major problem using artificial potential fields is the exhibition of local minima which could lead the robot to stack there before reaching its goal. The solution of the problem of finding an artificial potential field without local minima includes the employment of the so-called harmonic functions. Harmonic functions have indeed no local minima but are rather difficult to find for arbitrary boundary conditions. Therefore the Boundary Elements Method is employed which results in an approximation of an analytic solution of the desired artificial potential field.