Back to research
Benchmark Results
Action-Conditioned Visual Prediction for Robot Control
A world-model validation record for predicting visual consequences, comparing candidate actions and identifying possible collision risk.
Published January 19, 2026
BridgeV2WWorld ModelPredictive Control
Predict before execution
BridgeV2W converts robot configuration, trajectories and spatial coordinates into pixel-level dynamic masks that condition visual prediction.
Foresight and obstacle avoidance
Comparing candidate actions with predictive visual futures enables robots to make informed obstacle avoidance and path planning decisions before execution.