Research
Human-Robot Cognitive Alignment
Aligning a robot's representations and intentions with human expectations, so instructions land as they were meant.
For a robot to be a useful collaborator, its understanding of a task has to line up with the person’s. We investigate how to align robot representations, intentions and explanations with human expectations, so instructions are interpreted as they were meant and the robot’s behaviour stays legible.
Legibility matters in both directions: the person should be able to predict the robot, and the robot should notice when it has misread the person.