Human Inspired Grip-Release Technique for Robot-Human Handovers

Abstract

Fluent and natural robot human handovers are essential for human robot collaborative tasks. The robot's grip-release action is important for achieving this fluency. This paper describes an experimental study investigating interaction forces during grip-release in human-human handovers comprising of 13 participant pairs and a sensor embedded object. The results from this study were used to create a human inspired, data-driven strategy for robot grip-release technique in robot human handovers. This strategy was then evaluated alongside other techniques for grip-release in a robot human handovers experimentation study involving 20 participants. It was concluded that the data-driven strategy outperformed other strategies in getting natural handovers by faster grip-release for the sensor embedded object.