In human conversations, a glance or a smile can say more than words. Therefore, robots that adapt effectively and exhibit these cues can significantly enhance human-robot collaboration, making interactions feel more intuitive and human-like.
Riverside Research is an American independent research center for the benefit of the government and public interest. Their focus is on contributing to solving defense and intelligence challenges. Among their many areas of expertise, the organisation explores AI and Machine Learning, including advanced human-machine testing environments to answer early questions about human-machine teaming.
As part of their research, the team has developed a simple prompt-based approach that allows a Furhat robot to generate facial expressions and gaze shifts without prior task specific training. This enables a more natural interaction with individuals or groups. An initial prompt provides instructions to a pre-trained large language model, enabling the robot to naturally adjust its facial expressions and eye gaze during conversations. This adaptable and easily customisable approach can enhance the robot’s social presence by making human-robot interactions more fluid and lifelike.
Munje, Michael J., et al. "Team3 challenge: Tasks for multi-human and multi-robot collaboration with voice and gestures." Companion of the 2023 ACM/IEEE International Conference on Human-Robot Interaction. 2023. https://www.riversideresearch.org/insights/academic-papers/generative-facial-expressions-human-robot
Serfaty, Gabriel J., Virgil O. Barnard, and Joseph P. Salisbury. "Generative facial expressions and eye gaze behavior from prompts for multi-human-robot interaction." Adjunct Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology. https://www.riversideresearch.org/insights/academic-papers/challenge-tasks-robot-collaboration
