Inverse Kinematics Optimization for Robotic Manipulators: Exploiting Process-Based Redundancy

In industrial and collaborative environments, robotic manipulators are extensively used due to their precision and adaptability in performing complex tasks. Inverse kinematics is used to determine the joint configurations needed to achieve a target end-effector pose, which is given by both position and orientation. In redundant systems, where the number of degrees of freedom exceeds the minimum required for reaching the target, multiple inverse kinematics solutions exist. This redundancy admits additional optimization objectives, used to enhance the robot’s performance. The aim of the thesis project is to develop an IK algorithm that allows a robotic arm to reach a specified target, while optimizing additional factors. The end-effector goal configuration is composed by a full defined target position (x, y, z) and a partial defined target orientation: the angles on the x and y axes are fixed, while the angle on the z axis is left unconstrained. Rotations are usually treated with quaternions that represent a full orientation, making it difficult to isolate individual axes. To address this, the BioIK package from MoveIt is studied and used, and a customized goal class is developed, enforcing partial orientation constraints fixing two axes, but leaving rotation on the third axis unconstrained. This introduces a level of redundancy that allows the optimization of additional factors, such as avoiding joint limits and minimizing joint displacement. In addition to these goals, the final configuration must be collision-free. The proposed approach has been tested both in a simulated environment using ROS and MoveIt, and on a real Comau Racer-5 COBOT, to demonstrate its capability to achieve optimized collision-free inverse kinematics solutions. The results confirm its effectiveness in reaching the final target while optimizing movement, respecting kinematic constraints, and ensuring safe and feasible configurations.