A Method of Estimating an Object’s Parameters Based on Simplication with Momentum for a Manipulator

Abstract

For a manipulator to estimate an object’s inertial parameters—such as mass, center of mass position, and elements of the inertia tensor—it must grasp one side of the object and generate motion while utilizing the resulting forces and torques at its end-effector for estimation. In most previous studies, the estimation motion has involved high acceleration, resulting in larger motion trajectories and increased inertial forces. A larger trajectory raises the risk of collisions, while greater inertial forces could potentially damage the object. This paper introduces an innovative approach that simplifies the dynamic model using momentum, enabling accurate parameter estimation with minimal motion in robotic manipulation. Simulations are conducted to estimate key inertial parameters of the target object, including mass, center of mass position, and elements of the inertia tensor. The robotic manipulator securely grasps one side of the object and induces controlled motions to facilitate the estimation process. A comparative analysis with previously established estimation methods demonstrates that the proposed approach achieves accurate results with significantly smaller motions than prior techniques. The maximum acceleration reduction rates are 95% in linear motion and 98% in angular motion, respectively.