Atlas Grasping & Moving Objects, Trained on DextrAH-RGB

Robots are learning to grasp objects of various shapes and sizes. As AI advances, they get better at it. This video shows how Atlas was trained in simulation to grasp various objects. It pulls this off by relying on RL policies trained using NVIDIA’s DextrAH-RGB.

NVIDIA and Boston Dynamics have been working together to train and deploy DextrAH-RGB. The above video shows how a “robotic system driven by a generalist policy with robust, zero-shot simulation-to-real grasping capabilities” was deployed on Atlas’ upper torso.

[HT]