V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning

• URL: http://arxiv.org/abs/2506.09985v1
• Date: Wed, 11 Jun 2025 17:57:09 GMT
• Title: V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning
• Abstract summary: A major challenge for modern AI is to learn to understand the world and act largely by observation.<n>This paper explores a self-supervised approach that combines internet-scale video data with a small amount of interaction data.<n>We develop models capable of understanding, predicting, and planning in the physical world.
• Score: 43.18609951839598
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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.

Related papers

• AIM: Intent-Aware Unified world action Modeling with Spatial Value Maps [7.710034405765985]
  AIM is an intent-aware unified world action model that bridges this gap via an explicit spatial interface.<n>Built on a pretrained video generation model, AIM jointly models future observations and value maps within a shared mixture-of-transformers architecture.<n>Experiments on RoboTwin 2.0 benchmark show that AIM achieves a 94.0% average success rate, significantly outperforming prior unified world action baselines.
  arXiv  Detail & Related papers  (2026-04-13T07:48:58Z)
• GigaWorld-Policy: An Efficient Action-Centered World--Action Model [50.107640832046464]
  GigaWorld-Policy is an action-centered WAM that learns 2D pixel-action dynamics while enabling efficient action decoding, with optional video generation.<n> Experimental results on real-world robotic platforms show that GigaWorld-Policy runs 9x faster than the leading WAM baseline, Motus.
  arXiv  Detail & Related papers  (2026-03-18T00:52:02Z)
• Human-to-Robot Interaction: Learning from Video Demonstration for Robot Imitation [5.967530183571141]
  Human-to-Robot'' imitation learning pipeline enables robots to acquire manipulation skills directly from unstructured video demonstrations.<n>Key innovation is a modular framework that decouples the learning process into two distinct stages.<n>For robot manipulation, our framework achieves an average success rate of 87.5% across all actions, with 100% success on reaching tasks and up to 90% on complex pick-and-place operations.
  arXiv  Detail & Related papers  (2026-02-22T13:26:27Z)
• DreamDojo: A Generalist Robot World Model from Large-Scale Human Videos [110.98100817695307]
  We introduce DreamDojo, a foundation world model that learns diverse interactions and dexterous controls from 44k hours of egocentric human videos.<n>Our work enables several important applications based on generative world models, including live teleoperation, policy evaluation, and model-based planning.
  arXiv  Detail & Related papers  (2026-02-06T18:49:43Z)
• Large Video Planner Enables Generalizable Robot Control [117.49024534548319]
  General-purpose robots require decision-making models that generalize across diverse tasks and environments.<n>Recent works build robot foundation models by extending multimodal large language models (LMs) with action outputs, creating vision--action (VLA) systems.<n>We explore an alternative paradigm of using large-scale video pretraining as a primary modality for building robot foundation models.
  arXiv  Detail & Related papers  (2025-12-17T18:35:54Z)
• Scalable Vision-Language-Action Model Pretraining for Robotic Manipulation with Real-Life Human Activity Videos [42.86535655563404]
  We develop a fully-automated holistic human activity analysis approach for arbitrary human hand videos.<n>We process a large volume of egocentric videos and create a hand-VLA training dataset containing 1M episodes and 26M frames.<n>We design a dexterous hand VLA model architecture and pretrain the model on this dataset.
  arXiv  Detail & Related papers  (2025-10-24T15:39:31Z)
• AMPLIFY: Actionless Motion Priors for Robot Learning from Videos [29.799207502031496]
  We introduce AMPLIFY, a novel framework that leverages large-scale video data.<n>We train a forward dynamics model on abundant action-free videos and an inverse dynamics model on a limited set of action-labeled examples.<n>In downstream policy learning, our dynamics predictions enable a 1.2-2.2x improvement in low-data regimes, a 1.4x average improvement by learning from action-free human videos, and the first generalization to LIBERO tasks from zero in-distribution action data.
  arXiv  Detail & Related papers  (2025-06-17T05:31:42Z)
• HUMOTO: A 4D Dataset of Mocap Human Object Interactions [27.573065832588554]
  Human Motions with Objects is a high-fidelity dataset of human-object interactions for motion generation, computer vision, and robotics applications.<n>Humoto captures interactions with 63 precisely modeled objects and 72 articulated parts.<n>Professional artists rigorously clean and verify each sequence, minimizing foot sliding and object penetrations.
  arXiv  Detail & Related papers  (2025-04-14T16:59:29Z)
• Track2Act: Predicting Point Tracks from Internet Videos enables Generalizable Robot Manipulation [65.46610405509338]
  We seek to learn a generalizable goal-conditioned policy that enables zero-shot robot manipulation. Our framework,Track2Act predicts tracks of how points in an image should move in future time-steps based on a goal. We show that this approach of combining scalably learned track prediction with a residual policy enables diverse generalizable robot manipulation.
  arXiv  Detail & Related papers  (2024-05-02T17:56:55Z)
• Seer: Language Instructed Video Prediction with Latent Diffusion Models [43.708550061909754]
  Text-conditioned video prediction (TVP) is an essential task to facilitate general robot policy learning. We propose a sample and computation-efficient model, named textbfSeer, by inflating the pretrained text-to-image (T2I) stable diffusion models along the temporal axis. With the adaptable-designed architecture, Seer makes it possible to generate high-fidelity, coherent, and instruction-aligned video frames.
  arXiv  Detail & Related papers  (2023-03-27T03:12:24Z)
• RT-1: Robotics Transformer for Real-World Control at Scale [98.09428483862165]
  We present a model class, dubbed Robotics Transformer, that exhibits promising scalable model properties. We verify our conclusions in a study of different model classes and their ability to generalize as a function of the data size, model size, and data diversity based on a large-scale data collection on real robots performing real-world tasks.
  arXiv  Detail & Related papers  (2022-12-13T18:55:15Z)
• Masked World Models for Visual Control [90.13638482124567]
  We introduce a visual model-based RL framework that decouples visual representation learning and dynamics learning. We demonstrate that our approach achieves state-of-the-art performance on a variety of visual robotic tasks.
  arXiv  Detail & Related papers  (2022-06-28T18:42:27Z)
• Learning Generalizable Robotic Reward Functions from "In-The-Wild" Human Videos [59.58105314783289]
  Domain-agnostic Video Discriminator (DVD) learns multitask reward functions by training a discriminator to classify whether two videos are performing the same task. DVD can generalize by virtue of learning from a small amount of robot data with a broad dataset of human videos. DVD can be combined with visual model predictive control to solve robotic manipulation tasks on a real WidowX200 robot in an unseen environment from a single human demo.
  arXiv  Detail & Related papers  (2021-03-31T05:25:05Z)
• Learning Predictive Models From Observation and Interaction [137.77887825854768]
  Learning predictive models from interaction with the world allows an agent, such as a robot, to learn about how the world works. However, learning a model that captures the dynamics of complex skills represents a major challenge. We propose a method to augment the training set with observational data of other agents, such as humans.
  arXiv  Detail & Related papers  (2019-12-30T01:10:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.