Bridge Dataset V-2
Emerging7papers using it
2025first seen
The 'Bridge Dataset V2' is a benchmark that contains a variety of semantically equivalent but structurally diverse commands paired with existing trajectories, used to evaluate and enhance the linguistic generalization of Vision-Language-Action models like OpenVLA.
Papers using Bridge Dataset V-2 (5)
- Enhancing Linguistic Generalization of VLA: Fine-Tuning OpenVLA via Synthetic Instruction AugmentationMVP-LAM: Learning Action-Centric Latent Action via Cross-Viewpoint ReconstructionImage Generation as a Visual Planner for Robotic ManipulationDRAW2ACT: Turning Depth-Encoded Trajectories into Robotic Demonstration VideosSITCOM: Scaling Inference-time Compute For Vlas