MULTI-OBJECTIVE OPTIMIZATION FOR ENERGY-AWARE INVERSE KINEMATICS OF ROBOTICS

Inverse kinematics is the fundamental study of robot motion in space. Usually, the control command for a robot is only given by target position and orientation, in which the motions of the joint motors are found from the inverse kinematics. For the specified position or orientation, there are often multiple inverse kinematics solutions. The traditional method obtains all these solutions and then selects one of them according to some rules, such as minimum energy, shortest distance, shortest time, etc. In this thesis research, weighted multi-objective genetic algorithm (WMOGA) and weighted multi-objective particle swarm optimization (WMOPSO) are developed to solve energy-aware inverse kinematics by considering that the energy consumption of the robot is proportional to the total amount of joint rotations to the target pose. Therefore, the objective functions are proposed based on minimizing the target position error, target orientation error, and total amount of joint rotations to obtain the inverse kinematics solution. The algorithms developed in this thesis use a weighted sum of the multiple objective functions to iterate over a certain range of weights to obtain the inverse kinematics solution within the required error range. Furthermore, the proposed algorithm can be applied to the robotic model with any number of degrees of freedom. The trajectories solved by the proposed WMOGA and WMOPSO algorithms are verified by simulations and validated by the physical robotic arm. The achieved inverse kinematic solutions demonstrate a promising performance and have the potential for applications of the inverse kinematics of robotics.