Reasoning the Trust of Humans in Robots through Physiological Biometrics in Human-Robot Collaborative Contexts

With the rapid recent growth of automation and artificial intelligence, human-robot collaboration (HRC) is playing a significant role across a variety of fields. Trust between humans and robots is an important element to enable the efficiency and success of HRC. The lack of trust of humans in robots can have critical consequences, especially in real-world applications in which humans must adapt to unfamiliar situations. In this work, we develop a novel and effective approach for robots to actively reason and respond to dynamic human emotions and trust levels during shared tasks. We implement a real-world validation experiment in the context of human-robot object hand-over, which shows the robot’s ability to correctly identify and predict the human’s trust levels in real-time and assist the human accordingly in human-robot collaborative tasks. Future work on how to improve the performance of the proposed approach is also discussed.