Cross-embodiment learning
- In French:
- Apprentissage multi-robots
Definition
Training a single model on data from different robots, rather than a separate model for each robot, task and environment. The aim is a policy that adapts more easily to new robots and new tasks. To that end, the Open X-Embodiment project brought together a dataset collected on 22 robots by 21 institutions.
Checked on against the references below. · Written with AI · checked against the sources Source: Open X-Embodiment Collaboration, “Open X-Embodiment: Robotic Learning Datasets and RT-X Models”, 2023 (arXiv:2310.08864).