The library

Articles

Conceptual robot arm placing a wooden block on a stack; a faint path suggests adjustment between attempts.

How does reinforcement learning teach a robot through trial and feedback?

How task-specific rewards guide robot learning, illustrated by bounded physical walking and grasping studies.

Conceptual split view of a robot arm aligning a cylinder over a socket in simulation and on physical hardware.

Why can a robot trained in simulation fail in the real world?

Why simulator observations and physics can differ from hardware, with varied simulation, human correction and physical learning as research approaches.

Conceptual illustration of a robot arm, a displaced cube and suggested paths toward a target.

How do we test whether a robot can recover after an error?

How RoboRecover reconstructs simulated recovery points, compares clean-start and recovery results on LIBERO, and limits what the rates can show.