Co-VLA: Federated Training for Vision-Language-Action Models

Consensus-based federated training of robot VLA models without pooling robot data