What is robot learning?
Robot learning applies machine learning to perception, decision-making and control for physical agents. The output is not only a prediction; it can become a motor action with real consequences.
Why the problem is different
Robots face sensor noise, delayed feedback, contact dynamics, safety constraints and limited interaction data. A useful learner must account for these constraints while improving from demonstrations, simulation or experience.
Where the fields overlap
Robot learning uses supervised learning, imitation learning, reinforcement learning and representation learning. The difference is the physical grounding of the task and the need to evaluate behavior in an environment.
Explore the field
Continue through the research map, model directory, robot directory and company directory.