Monday 03 March 2025
As robots become increasingly common in our daily lives, it’s essential to understand how humans interact with them and what makes those interactions successful or unsuccessful. A recent study published in a scientific journal has shed light on this topic by examining the relationship between robot competence, motion legibility, and human correction feedback.
The researchers recruited 60 participants who were asked to supervise and correct a robot performing pick-and-place tasks. The robot was programmed to perform differently depending on its level of competence and the legibility of its motions. Legibility refers to how easy it is for humans to understand what the robot is trying to do, based on its movements.
The results showed that when the robot performed well but its motions were hard to follow, participants were more likely to correct its mistakes. On the other hand, when the robot struggled with the task and its motions were clear, participants were less likely to intervene. This suggests that humans are more willing to help a competent robot that’s having trouble because they can see what it’s trying to do.
The study also found that as the robot’s competence increased, participants became more sensitive to suboptimal behavior. In other words, if the robot was performing well but made small mistakes, participants were more likely to correct them. This implies that humans have high expectations for robots that are capable of complex tasks and are willing to provide feedback to help them improve.
One interesting finding was that physical effort positively correlated with correction precision. This means that when participants had to exert more effort to guide the robot towards the correct goal, they were able to achieve higher levels of accuracy. However, this correlation was weaker for incompetent robots with legible motions, suggesting that humans may be less motivated to help a robot that’s struggling if it’s not clear what it’s trying to do.
The study’s findings have implications for designing robot interaction behaviors and learning task objectives from corrections. By taking into account the relationship between robot competence, motion legibility, and human correction feedback, developers can create more effective and efficient interactions between humans and robots.
For example, if a robot is designed to perform complex tasks but its motions are unclear, it may be beneficial to incorporate additional visual or auditory cues to help humans better understand what it’s trying to do. Similarly, if a robot is struggling with a task, developers could design the interface to provide more explicit feedback to participants, encouraging them to intervene and correct the robot’s mistakes.
Cite this article: “Understanding Human-Robot Interaction: The Role of Competence, Legibility, and Feedback”, The Science Archive, 2025.
Robots, Human Interaction, Competence, Motion Legibility, Correction Feedback, Pick-And-Place Tasks, Robot Performance, Participant Behavior, Design Implications, Interface Development
