Admittance Control for Reducing Human Physical Effort in Robot Programming through Motion Guidance
摘要
This paper describes an admittance control in collaborative robots (cobots) to reduce the physical effort required from operators during the manual guidance of the cobot’s end-effector to program linear trajectories. This technique, known as programming by demonstration, allows operators to physically guide the cobot’s end-effector, teaching it a desired motion pattern. The control strategy proposed here is tailored for industrial applications and offers a significant safety advantage by avoiding direct manipulation of the joint torques, which often cannot be accessed in industrial robots due to safety regulations. Instead, our approach measures the force exerted by the user via a piezoresistive FSR402 sensor located on the cobot’s end-effector, and converts this force into a velocity set point through a second-order mass-spring-damper system implemented in software, which represents the desired admittance model. This velocity then serves as the input for the robot’s existing velocity control algorithms. In this study, we analyze the effect of the mass of the desired admittance model on the force exerted by four volunteers using a UR3 cobot. The experiments also involved a second UR3 cobot used to emulate mechanical perturbations affecting the cobot moved by the human operator. Statistical analyses using ANOVA and Tukey’s Honestly Significant Differences tests were conducted under two distinct conditions: with and without mechanical perturbations. These tests confirmed that admittance control with a