V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning
About
A major challenge for modern AI is to learn to understand the world and learn to act largely by observation. This paper explores a self-supervised approach that combines internet-scale video data with a small amount of interaction data (robot trajectories), to develop models capable of understanding, predicting, and planning in the physical world. We first pre-train an action-free joint-embedding-predictive architecture, V-JEPA 2, on a video and image dataset comprising over 1 million hours of internet video. V-JEPA 2 achieves strong performance on motion understanding (77.3 top-1 accuracy on Something-Something v2) and state-of-the-art performance on human action anticipation (39.7 recall-at-5 on Epic-Kitchens-100) surpassing previous task-specific models. Additionally, after aligning V-JEPA 2 with a large language model, we demonstrate state-of-the-art performance on multiple video question-answering tasks at the 8 billion parameter scale (e.g., 84.0 on PerceptionTest, 76.9 on TempCompass). Finally, we show how self-supervised learning can be applied to robotic planning tasks by post-training a latent action-conditioned world model, V-JEPA 2-AC, using less than 62 hours of unlabeled robot videos from the Droid dataset. We deploy V-JEPA 2-AC zero-shot on Franka arms in two different labs and enable picking and placing of objects using planning with image goals. Notably, this is achieved without collecting any data from the robots in these environments, and without any task-specific training or reward. This work demonstrates how self-supervised learning from web-scale data and a small amount of robot interaction data can yield a world model capable of planning in the physical world.
Key insights
What problem does V-JEPA 2 address?
V-JEPA 2 learns to understand, predict, and plan in the physical world by self-supervised pretraining on over 1 million hours of internet video, then post-training an action-conditioned world model on 62 hours of unlabeled robot data.
How does V-JEPA 2 differ from prior work?
Unlike video generation models, V-JEPA 2 predicts in representation space, ignoring unpredictable pixel details. It achieves 77.3 top-1 on SSv2 and 39.7 recall-at-5 on Epic-Kitchens-100, a 44% relative improvement over the previous best.
What are the practical findings useful for?
V-JEPA 2-AC was deployed zero-shot on Franka arms in two different labs for prehensile manipulation tasks like pick-and-place, using planning with image goals, without collecting environment-specific data or task-specific rewards.
What limitations do the authors state?
The authors note V-JEPA 2-AC is sensitive to camera positioning, suffers error accumulation in long-horizon planning, and currently relies on image goals rather than language, which they propose addressing in future work.