Article

Ground-JEPA: Learning Physically Grounded Latent World Models for Zero-Shot Dynamics Generalization of Legged Robots

2026-07-29

Abstract excerpt

<title>Abstract</title> <p> Accurate dynamics models are fundamental to model-based control of legged robots, yet existing neural network approaches suffer from autoregressive error accumulation over long horizons. Joint-Embedding Predictive Architectures (JEPAs) offer an alternative by learning dynamics in a compact latent space without reconstructing observations. Here we introduce Ground-JEPA, a minimal (~1.3...

Topics

Open a Topic to create a Post that cites this publication.

Identifiers and source

Literature Corpus work: bbe69cbe-0daf-5b30-b811-06c05e9876ac
DOI: 10.21203/rs.3.rs-10511499/v1

Open publication

Related research

Semantic proximity does not establish scientific evidence.

Click a neighbor to travelStep 1 · 12 closest

Ground-JEPA: Learning Physically Grounded Latent World Models for Zero-Shot Dynamics Generalization of Legged RobotsDOI 10.21203/rs.3.rs-10511499/v1Read

Lifelike Agility and Play on Quadrupedal Robots using Reinforcement Learning and Generative Pre-trained ModelsSemantic proximity Read

Beyond Kinesthetic Twins: A Dematerialized Control Primitive for Zero-Shot Generalization Across Robot MorphologiesSemantic proximity Read

Deep kinematic inference affords efficient and scalable control of bodily movementsSemantic proximity Read

iLQR-VAE : control-based learning of input-driven dynamics with applications to neural dataSemantic proximity Read

Synthetic Data-Driven Exoskeleton Control via Contralateral Gait Fusion for Variable-Speed WalkingSemantic proximity Read

EnerVerse-AC: Envisioning Embodied Environments with Action ConditionSemantic proximity Read

Physical AI: The Next Frontier in AI and Robotics to Build Truly Autonomous MachinesSemantic proximity Read

Model-Based Reinforcement Learning for Ultrasound-Driven Autonomous MicrorobotsSemantic proximity Read

Bidirectional Dynamic Neural Networks with Physical AnalyzabilitySemantic proximity Read

Environment Classification for Robotic Leg Prostheses and Exoskeletons using Deep Convolutional Neural NetworksSemantic proximity Read

Reward-predictive representations generalize across tasks in reinforcement learningSemantic proximity Read

Vision-Language-Action and Vision Language Models for Robot Manipulation: A Comprehensive Review Towards Real-World ApplicationsSemantic proximity Read