Hough transform for robot motion planning in industrial applications

This paper introduces a novel technique for robot motion planning by using the Hough transform. Although path planning is computed in a two-dimensional configuration space, both mobile robot and obstacles are modelled by 3D geometrical structures. Obstacle avoidance is based on the computation of the translational distance of penetration. The proposed penetration distance allows the planner to obtain a point of the path where the obstacle is optimally avoided. Each obstacle is then represented by such a point. The Hough transform applied to the points detects what obstacles are essential to be avoided in order to generate an optimal collision-free path. The computational cost of the path planner is linear with the number of obstacles. The collision detection between two objects is determined by applying the Hough transform and computing the relative position of the origin with respect to the Minkowski difference of the two objects. Hough transform has also been used by the authors for automatic generation of geometric models