Bridge V-2

'Bridge V-2' is a benchmark dataset that contains time-synchronized multi-view videos used to evaluate the effectiveness of learning action-centric latent actions in robot learning and vision-language-action model pretraining.

Papers using Bridge V-2 (6)

Enhancing Linguistic Generalization of VLA: Fine-Tuning OpenVLA via Synthetic Instruction Augmentation2026MVP-LAM: Learning Action-Centric Latent Action via Cross-Viewpoint Reconstruction2026Image Generation as a Visual Planner for Robotic Manipulation2025DRAW2ACT: Turning Depth-Encoded Trajectories into Robotic Demonstration Videos2025A Taxonomy for Evaluating Generalist Robot Manipulation Policies2025Action-Free Reasoning for Policy Generalization2025