A Comparative Study for Obstacle Avoidance Inverse Kinematics: Null-Space Based vs. Optimisation-Based

Obstacle avoidance for robotic manipulators has attracted much attention in robotics literature, and many algorithms have been developed. In this paper, two algorithms are explored which perform obstacle avoidance within the inverse kinematics calculation process. The first algorithm applies a velocity to the null-space of the manipulator’s Jacobian matrix which directs the manipulator away from obstacles. The second algorithm uses an optimisation-based approach to calculate joint positions which incorporates constraints to prevent the manipulator from violating obstacle boundaries. Applying obstacle avoidance at the inverse kinematics level of the control process is particularly applicable to teleoperation and allows the robotic manipulator to react to obstacles at a faster rate without involving a path planner which operates on a slower cycle time. The two algorithms were implemented for a direct comparison in terms of obstacle avoidance capability and processing times. It was found that the processing time of the null-space method was substantially quicker than the optimisation-based algorithm. However, the null-space method did not guarantee collision avoidance which may not be suitable for safety critical applications without supervision.