GENERIK: An Objective function Optimization Approach to Inverse Kinematics

Abstract

This paper covers the conceptualization, algorithmic procedure, and implementation of a novel inverse kinematic solver named GENERIK. Here inverse kinematics is formulated as an optimization problem and the solution is obtained by minimizing the objective function. A study is conducted choosing “Mean Squared Error (MSE)” and “Logarithm of the Hyperbolic Cosine (Logcosh)” as the objective functions and the resulting error, performance, and other trade-offs are reported. GENERIK has been validated using “Kuka LBR iiwa” (7DOF), “UR5” (6DOF), and “YouBot Arm” (5DOF), on Robot Operating System (ROS) supported simulators such as Rviz. Further, the robotic manipulator (6DOF) of “ARDOP: A Humanoid Robotic Research Platform” is used to evaluate the performance of GENERIK in hardware.