Kim, Hyeongjun (KAIST), Park, Jeongsu (KAIST), Youn, Jimin (Korea Advanced Institute of Science and Technology (KAIST)), Kim, Jongwon (KAIST), Kong, Kyoungchul (Korea Advanced Institute of Science and Technology)

Dynamic Modeling of Geared Actuator and Model Parameter Identification Method in Assembled Robots

Scheduled for presentation during the Regular Session "Modeling, Identification, and Estimation" (WeCT2), Wednesday, July 16, 2025, 17:30−17:50, Room 106

Joint 10th IFAC Symposium on Mechatronic Systems and 14th Symposium on Robotics, July 15-18, 2025, Paris, France

This information is tentative and subject to change. Compiled on July 16, 2025

Abstract

Control algorithms based on dynamic models of robotic systems, such as reinforcement learning and Model Predictive Control (MPC), have become central to motion control. Achieving reliable sim-to-real transfer necessitates precise modeling and parameter identification, including for actuators. Conventional actuator models typically consider only motion states, neglecting load dependent friction variations. This issue becomes more pronounced as the gear ratio increases and the torque transmitted through the gear rises. This study presents a dynamic modeling approach accounting for load effects and proposes an experimental parameter identification method conducted on assembled robots, to enhance the accuracy of identification and eliminate the need for complex setups. The experimental results demonstrate that using the proposed model and its parameters allows for a more precise depiction of actuator behavior.