A Multipurpose VR-Based Skill-Transfer for Robotic Platform
摘要
This paper presents a novel VR-Based Robotic Skill-Transfer System called LinkForm. Our system comprises two synchronized VR environments for the trainee and coach, cameras, and a robot arm that the coach controls to evaluate the posture and guide the hand movements of the trainee. Besides visual and verbal instructions in the VR environment, LinkForm allows physical guidance of the trainee’s posture and hand by the couch, which can greatly improve skill transfer for scenarios requiring precise hand motions or specific body postures. To explore the potential of LinkForm, we co-designed and implemented six training scenarios with experts. Furthermore, we evaluated LinkForm’s skill-transfer capability with 12 participants in a real table tennis training scenario. Our results show statistically significant skill improvement after training with LinkForm, and participants had generally positive feedback of LinkForm. Lastly, we discuss the future prospects of LinkForm in light of our design and evaluation.