SyncArms: Gaze-driven implicit assistive control for parallel robot arms in distributed environments

Author

Kawamura, Koki and Kasahara, Shunichi and Sugimoto, Maki

Abstract

Enhancing the human capability of a single person by using multiple bodies has been a significant research agenda for human augmentation. When multiple bodies are synchronously operated in different environments, the differences in environments make it difficult to interact with objects simultaneously, and the task performance is lowered because some tasks cannot be performed in parallel. Therefore, an automatic assistive interface is required to perform these tasks in parallel. However, the motion of automatic control generates visuo-motor incongruences resulting in a loss of embodiment over the multiple bodies. Thus, absorbing differences in placement to perform parallel tasks and increasing task performance involves the trade-off of a reduced embodiment. To address this trade-off between task performance and embodiment, we have developed a system that allows a parallel operation with synchronized multiple robotic arms while maintaining a sense of embodiment. This is achieved through detecting the target and environment that the user is currently observing by eye tracking and object recognition, and implicitly providing positional assistance to the robotic arm in the unobserved environment to interact with the corresponding target. To evaluate our system, we conducted an experiment to compare it with existing methods. The results show that the proposed method could provide the comparable embodiment as when all arms are completely synchronized, and the comparable task performance as when automatic control is performed regardless of whether the environment is aligned or not. Also, negative correlations were found between the embodiment and whether the user was aware of the automatic assistive controls. We believe that our findings provide benefits of maintaining user’s embodiment when designing an interface for operating multiple bodies in parallel.