Incorporating Implicit Human Feedback to Better Learn from Explicit Human Feedback in Human-Robot Interactions

Our goal is to enhance the ability of robots to learn how to assist people by incorporating implicit human feedback during human-robot interactions. Since the way individuals prefer to receive assistance from robots can be influenced by various factors, it is crucial for robots to adapt their behavior based on a person’s changing preferences during an interaction. Conventionally, robots acquire knowledge from humans through direct feedback, such as evaluations, preferences, demonstrations, or corrections. However, this kind of feedback can disrupt the natural flow of interaction. Since humans unconsciously convey information through non-verbal cues, we aim to incorporate this implicit feedback to improve how robots learn from humans. We are investigating different methods for incorporating the feedback that humans convey implicitly into robot learning processes. By intelligently integrating explicit and implicit human feedback, our aim is to make human-robot interactions more seamless and intuitive for humans.