A robotic hand that utilizes ergonomic evaluation as feedback to improve human robot collaboration in soldering applications

Abstract

People never seem to have enough hands. There are many tools that aim to address this challenge, ranging from the ubiquitous benchtop vise to the "helping hands" commonly used for soldering. However, these tools do not measure up to their human counterparts. They cannot adjust the position or orientation of the workpiece to suit a particular task which can cause workers to maintain unhealthy postures that are detrimental to their long-term health. This thesis addresses this shortcoming with a robotic arm that utilizes a gripper to grasp and hold a workpiece during a soldering task. The robot uses a Microsoft Kinect sensor to continuously analyze the posture of the human worker and calculate a score based on the RULA (Rapid Upper Limb Assessment), an objective measure used in the ergonomics field to evaluate ergonomic working postures. The robot adjusts the workpiece in order to optimize the RULA score using an adaptive simulated annealing algorithm to balance the exploration and exploitation phases of the optimization process. Initial testing indicates that the robot can consistently find positions which improve the RULA ranking by 24.6% of the measured range. This project demonstrates that human robot collaboration can be improved by utilizing sensors to evaluate the needs of a human partner and adjust the robot behavior accordingly.