In this paper, an on-line path planner for an industrial manipulator is presented. The proposed control architecture is capable of driving the manipulator in its environment while avoiding collisions. Potential fields are used in order to control the joint velocities in such a way that the robot avoids the obstacles. We also propose a new weighted pseudoinverse matrix that improves the manipulator capability of finding feasible paths to move around obstacles and pass through narrow corridors without relying on the manipulator dynamic model. The proposed technique fits to both redundant and non-redundant manipulators. Experimental results show the effectiveness of the proposed solution.