Villa Xenia

New LAWM framework enables self-supervised robotic learning from video

Researchers have introduced LAWM, a novel framework for self-supervised pretraining of imitation learning models in robotics. This model-agnostic approach learns latent action representations from unlabeled video data b…