RT-1
Robotics Transformer 1 and its 130k-episode dataset of real-world tasks. · By Google
RT-1 is Google's transformer policy trained on 130k demonstrations over 17 months on 13 robots, covering 700+ tasks. The dataset is part of Open X-Embodiment and the model code
Description
RT-1 is Google's transformer policy trained on 130k demonstrations over 17 months on 13 robots, covering 700+ tasks. The dataset is part of Open X-Embodiment and the model code is open.