Development and evaluation of zoomorphic gestural communication on a spot quadruped robot

摘要

Bi-lateral communication between humans and robots is critical to their effective collaboration and the successful adoption of robotics in industry and as consumer products by the general public. While significant work has been done into communicating human intent to robots, less research has investigated conveying robotic intent to people. This study investigates the use of zoomorphic gestural communication on a commercially available Boston Dynamics Spot quadruped platform to determine if people can reliably recognise the robot’s intended actions from its movement. This is of particular interest as legged platforms bear much closer resemblance to living creatures than wheeled platforms or robotic arms and have not been thoroughly explored in research before. Specifically, we investigate whether ‘gestural building blocks’, based on the existing movement capabilities of the robot, can be composed together to successfully convey a range of intended gestural meanings to a person. To assess this, videos of four distinct gestural movement sequences related to a workplace inspection scenario were presented to participants who then completed a multi-choice survey and the Godspeed indices to measure their interpretation of each gesture and their overall impression of the robot across the key areas of anthropomorphism, likeability, perceived intelligence and perceived safety. Our results show that zoomorphic gestural communication using the existing gestural abilities of a quadruped robot produced successfully identified gestural meaning the majority of the time for all videos and that participants overall held positive attitudes towards the robot. This finding provides the groundwork for further exploration of the gestural communication modality for robots of this class.