Human-to-Robot Imitation in the Wild
Shikhar Bahl, Abhinav Gupta, Deepak Pathak
Abstract
We present WHIRL, an efficient real-world algorithm for one-shot visual imitation in the wild. WHIRL is able to directly learn from unstructured human videos and generalize to new tasks as well. Videos and webpage at: https://human2robot.github.io
Topics & Concepts
ImitationRobotComputer scienceHuman–robot interactionArtificial intelligenceComputer visionHuman–computer interactionPsychologyNeuroscienceRobotics and Automated SystemsRobot Manipulation and LearningRobotic Locomotion and Control