Processed egocentric human video outperforms teleoperated real-robot trajectories as pretraining data for embodied foundation models, delivering 24% lower validation loss and 52.5-90% higher task success rates under matched post-training protocols.
hub Canonical reference
R3M: A Universal Visual Representation for Robot Manipulation
Canonical reference. 93% of citing Pith papers cite this work as background.
62
Pith papers citing it
82
external citations · Pith
Background
93% of classified citations
abstract
We study how visual representations pre-trained on diverse human video data can enable data-efficient learning of downstream robotic manipulation tasks. Concretely, we pre-train a visual representation using the Ego4D human video dataset using a combination of time-contrastive learning, video-language alignment, and an L1 penalty to encourage sparse and compact representations. The resulting representation, R3M, can be used as a frozen perception module for downstream policy learning. Across a suite of 12 simulated robot manipulation tasks, we find that R3M improves task success by over 20% compared to training from scratch and by over 10% compared to state-of-the-art visual representations like CLIP and MoCo. Furthermore, R3M enables a Franka Emika Panda arm to learn a range of manipulation tasks in a real, cluttered apartment given just 20 demonstrations. Code and pre-trained models are available at https://tinyurl.com/robotr3m.
hub tools
citation-role summary
background 13
method 1
citation-polarity summary
representative citing papers
C4G introduces compact timestamp-conditioned Gaussian query tokens that aggregate full temporal context to decode 3D Gaussians with timestamp-modulated positions for feed-forward 4D reconstruction from monocular video, plus a diffusion-based rendering module and extension to 4D feature fields.
JOPAT jointly models pixels, point tracks, and actions in a diffusion transformer and reports gains over pixel-only baselines on long-horizon robot tasks with occlusion and off-screen motion.
RoboWM-Bench evaluates video world models by converting their manipulation video predictions into executable actions validated in simulation, showing that visual plausibility does not guarantee physical executability.
Multimodal Diffusion Forcing trains a diffusion model on partially masked multimodal robot trajectories to learn temporal and cross-modal dependencies for forceful manipulation.
DreamGen trains robot policies on synthetic trajectories from adapted video world models, enabling a humanoid robot to perform 22 new behaviors in seen and unseen environments from a single pick-and-place teleoperation dataset.
SuSIE uses a finetuned InstructPix2Pix diffusion model to propose subgoal images that guide a low-level goal-conditioned policy, achieving SOTA zero-shot performance on CALVIN and real-world manipulation.
VoxPoser uses LLMs to compose 3D value maps via VLM interaction for model-based synthesis of robust robot trajectories on open-set language-specified manipulation tasks.
VIP learns a visual embedding from human videos whose distance defines dense, smooth rewards for arbitrary goal-image robot tasks without task-specific fine-tuning.
SayCan combines an LLM's high-level semantic knowledge with robot skill value functions to select only feasible actions, enabling completion of abstract natural-language instructions on a real mobile manipulator.
World Action Model co-training with DINO or 3D-flow targets scales human-to-robot transfer on bimanual tasks far better than behavior cloning, while pixel prediction transfers weakly.
STEAM learns advantages from expert trajectories via self-supervised temporal ensemble modeling to improve policy learning on real robot tasks like bimanual folding and pick-and-place.
GRA extracts 2D waypoints from synthetic videos to supervise VLA vision while restricting action training to real data, outperforming pseudo-action baselines on real-robot tasks.
OpenHLM is an empirical recipe yielding a whole-body humanoid VLA model that outperforms GR00T N1.6 and Ψ0 baselines on long-horizon tasks using less than half the demonstration time.
GLAM learns a shared latent action space grounded in consistent future observation prediction across heterogeneous data sources to train improved behavioral cloning policies for robot manipulation tasks.
PoLAR imposes radial structure on latent actions in hyperbolic space to factorize extent and mode, improving robot policy performance over baselines.
DO AS I DO reconstructs and retargets hand-object interactions from in-the-wild monocular RGB videos to produce dexterous robot manipulation trajectories, outperforming prior methods on ground-truth and online video datasets.
EgoInfinity is a modular pipeline that lifts in-the-wild RGB videos into agent-agnostic 4D hand-object data with interaction-aware refinement and retargets motions to diverse robot morphologies for video-to-action learning.
CAIP learns action-aligned visual representations via contrastive pre-training on human hand keypoints from egocentric video, outperforming DINOv2, SigLIP, MVP, and R3M with >30% gains on real dexterous manipulation tasks.
Introduces Embodied Tool Protocol and tool externalization to improve embodied AI performance on perception and cognition tasks, with measured gains but limits on execution capabilities.
DeMiAn re-annotates robot and egocentric videos with VLM-generated dense labels across motion, scene, pose, and reasoning aspects, then uses a learned instructor to boost policy success by 5 points on RoboCasa over task-only baselines.
LACE aligns human-robot visual features via semantic distribution matching on corresponding body parts plus Gram loss, yielding 65% better zero-shot policy transfer than baseline DINO.
GuidedVLA improves VLA generalization by supervising individual attention heads with manually defined auxiliary signals for three task-relevant factors.
HumanNet is a 1M-hour human-centric video dataset with interaction annotations that enables better vision-language-action model performance than equivalent robot data in a controlled test.
citing papers explorer
-
HumanScale: Egocentric Human Video Can Outperform Real-Robot Data for Embodied Pretraining
Processed egocentric human video outperforms teleoperated real-robot trajectories as pretraining data for embodied foundation models, delivering 24% lower validation loss and 52.5-90% higher task success rates under matched post-training protocols.
-
Learning Global Motion with Compact Gaussians for Feed-Forward 4D Reconstruction
C4G introduces compact timestamp-conditioned Gaussian query tokens that aggregate full temporal context to decode 3D Gaussians with timestamp-modulated positions for feed-forward 4D reconstruction from monocular video, plus a diffusion-based rendering module and extension to 4D feature fields.
-
Point Tracking Improves World Action Models
JOPAT jointly models pixels, point tracks, and actions in a diffusion transformer and reports gains over pixel-only baselines on long-horizon robot tasks with occlusion and off-screen motion.
-
RoboWM-Bench: A Benchmark for Evaluating World Models in Robotic Manipulation
RoboWM-Bench evaluates video world models by converting their manipulation video predictions into executable actions validated in simulation, showing that visual plausibility does not guarantee physical executability.
-
Multimodal Diffusion Forcing for Forceful Manipulation
Multimodal Diffusion Forcing trains a diffusion model on partially masked multimodal robot trajectories to learn temporal and cross-modal dependencies for forceful manipulation.
-
DreamGen: Unlocking Generalization in Robot Learning through Video World Models
DreamGen trains robot policies on synthetic trajectories from adapted video world models, enabling a humanoid robot to perform 22 new behaviors in seen and unseen environments from a single pick-and-place teleoperation dataset.
-
Zero-Shot Robotic Manipulation with Pretrained Image-Editing Diffusion Models
SuSIE uses a finetuned InstructPix2Pix diffusion model to propose subgoal images that guide a low-level goal-conditioned policy, achieving SOTA zero-shot performance on CALVIN and real-world manipulation.
-
VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models
VoxPoser uses LLMs to compose 3D value maps via VLM interaction for model-based synthesis of robust robot trajectories on open-set language-specified manipulation tasks.
-
VIP: Towards Universal Visual Reward and Representation via Value-Implicit Pre-Training
VIP learns a visual embedding from human videos whose distance defines dense, smooth rewards for arbitrary goal-image robot tasks without task-specific fine-tuning.
-
Do As I Can, Not As I Say: Grounding Language in Robotic Affordances
SayCan combines an LLM's high-level semantic knowledge with robot skill value functions to select only feasible actions, enabling completion of abstract natural-language instructions on a real mobile manipulator.
-
EgoWAM: World Action Models Beyond Pixels with In-the-Wild Egocentric Human Data
World Action Model co-training with DINO or 3D-flow targets scales human-to-robot transfer on bimanual tasks far better than behavior cloning, while pixel prediction transfers weakly.
-
STEAM: Self-Supervised Temporal Ensemble Advantage Modeling for Real-World Robot Learning
STEAM learns advantages from expert trajectories via self-supervised temporal ensemble modeling to improve policy learning on real robot tasks like bimanual folding and pick-and-place.
-
Supervise What Survives: Geometry-Guided VLA Adaptation from Synthetic Robot Videos
GRA extracts 2D waypoints from synthetic videos to supervise VLA vision while restricting action training to real data, outperforming pseudo-action baselines on real-robot tasks.
-
OpenHLM: An Empirical Recipe for Whole-Body Humanoid Loco-Manipulation
OpenHLM is an empirical recipe yielding a whole-body humanoid VLA model that outperforms GR00T N1.6 and Ψ0 baselines on long-horizon tasks using less than half the demonstration time.
-
Imitation from Heterogeneous Demonstrations using Grounded Latent-Action World Models
GLAM learns a shared latent action space grounded in consistent future observation prediction across heterogeneous data sources to train improved behavioral cloning policies for robot manipulation tasks.
-
PoLAR: Factorizing Extent and Mode in Latent Actions for Robot Policy Learning
PoLAR imposes radial structure on latent actions in hyperbolic space to factorize extent and mode, improving robot policy performance over baselines.
-
Do as I Do: Dexterous Manipulation Data from Everyday Human Videos
DO AS I DO reconstructs and retargets hand-object interactions from in-the-wild monocular RGB videos to produce dexterous robot manipulation trajectories, outperforming prior methods on ground-truth and online video datasets.
-
EgoInfinity: A Web-Scale 4D Hand-Object Interaction Data Engine for Any-View Robot Retargeting and Video-to-Action Robot Learning
EgoInfinity is a modular pipeline that lifts in-the-wild RGB videos into agent-agnostic 4D hand-object data with interaction-aware refinement and retargets motions to diverse robot morphologies for video-to-action learning.
-
Contrastive Action-Image Pre-training for Visuomotor Control
CAIP learns action-aligned visual representations via contrastive pre-training on human hand keypoints from egocentric video, outperforming DINOv2, SigLIP, MVP, and R3M with >30% gains on real dexterous manipulation tasks.
-
Enabling Extensible Embodied Capabilities with Tools
Introduces Embodied Tool Protocol and tool externalization to improve embodied AI performance on perception and cognition tasks, with measured gains but limits on execution capabilities.
-
How to Instruct Your Robot: Dense Language Annotations Power Robot Policy Learning
DeMiAn re-annotates robot and egocentric videos with VLM-generated dense labels across motion, scene, pose, and reasoning aspects, then uses a learned instructor to boost policy success by 5 points on RoboCasa over task-only baselines.
-
LACE: Latent Visual Representation for Cross-Embodiment Learning
LACE aligns human-robot visual features via semantic distribution matching on corresponding body parts plus Gram loss, yielding 65% better zero-shot policy transfer than baseline DINO.
-
GuidedVLA: Specifying Task-Relevant Factors via Plug-and-Play Action Attention Specialization
GuidedVLA improves VLA generalization by supervising individual attention heads with manually defined auxiliary signals for three task-relevant factors.
-
HumanNet: Scaling Human-centric Video Learning to One Million Hours
HumanNet is a 1M-hour human-centric video dataset with interaction annotations that enables better vision-language-action model performance than equivalent robot data in a controlled test.
-
Decompose and Recompose: Reasoning New Skills from Existing Abilities for Cross-Task Robotic Manipulation
Decompose and Recompose decomposes seen robotic demonstrations into skill-action alignments and recomposes them via visual-semantic retrieval and planning to enable zero-shot cross-task generalization.
-
GazeVLA: Learning Human Intention for Robotic Manipulation
GazeVLA pretrains on large human egocentric datasets to capture gaze-based intention, then finetunes on limited robot data with chain-of-thought reasoning to achieve better robotic manipulation performance than baselines.
-
UniT: Toward a Unified Physical Language for Human-to-Humanoid Policy Learning and World Modeling
UniT creates a unified physical language via visual anchoring and tri-branch reconstruction to enable scalable human-to-humanoid transfer for policy learning and world modeling.
-
WARPED: Wrist-Aligned Rendering for Robot Policy Learning from Egocentric Human Demonstrations
WARPED synthesizes realistic wrist-view observations from monocular egocentric human videos via foundation models, hand-object tracking, retargeting, and Gaussian Splatting to train visuomotor policies that match teleoperation success rates on five tabletop tasks with 5-8x less collection effort.
-
ProGAL-VLA: Grounded Alignment through Prospective Reasoning in Vision-Language-Action Models
ProGAL-VLA uses 3D graphs, symbolic sub-goals, and a Grounding Alignment Contrastive loss to ground actions on verified embeddings, raising robustness from 30.3% to 71.5% and ambiguity AUROC to 0.81 on robotic benchmarks.
-
PALM: Progress-Aware Policy Learning via Affordance Reasoning for Long-Horizon Robotic Manipulation
PALM improves long-horizon robotic manipulation success by distilling affordance representations for object interaction and predicting within-subtask progress in a VLA model.
-
Learning to Feel the Future: DreamTacVLA for Contact-Rich Manipulation
A vision-language-action policy that predicts future tactile images and uses that predicted touch to refine its actions reaches up to 95% success on contact-rich manipulation.
-
Goal-Driven Reward by Video Diffusion Models for Reinforcement Learning
Video diffusion models supply goal-driven rewards for RL by measuring alignment of agent trajectories with generated goal videos at both video and frame levels.
-
Uni-Hand: Universal Hand Motion Forecasting in Egocentric Views
Uni-Hand forecasts 2D/3D hand waypoints, head motion, and contact states in egocentric views using vision-language fusion and dual-branch diffusion, with new benchmarks for downstream robotics and action tasks.
-
InternVLA-M1: A Spatially Guided Vision-Language-Action Framework for Generalist Robot Policy
InternVLA-M1 uses spatially guided pre-training on 2.3M examples followed by action post-training to deliver up to 17% gains on robot manipulation benchmarks and 20.6% on unseen objects.
-
EgoVLA: Learning Vision-Language-Action Models from Egocentric Human Videos
EgoVLA pretrains VLA models on egocentric human videos, retargets predicted actions to robots via IK, and fine-tunes on few robot demos to improve bimanual manipulation performance on a new simulation benchmark.
-
DexWild: Dexterous Human Interactions for In-the-Wild Robot Policies
DexWild co-trains dexterous robot policies on in-the-wild human hand interactions recorded with a low-cost system and limited robot data, achieving 68.5% success in unseen environments and 5.8x better cross-embodiment generalization.
-
GR00T N1: An Open Foundation Model for Generalist Humanoid Robots
GR00T N1 is a new open VLA foundation model for humanoid robots that outperforms imitation learning baselines in simulation and shows strong performance on real-world bimanual manipulation tasks.
-
Video Prediction Policy: A Generalist Robot Policy with Predictive Visual Representations
Video Prediction Policy conditions robot action learning on future-frame predictions inside fine-tuned video diffusion models, yielding 18.6% relative gains on Calvin ABC-D and 31.6% higher real-world success rates.
-
DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planning
DINO-WM builds world models on pre-trained DINOv2 features to enable zero-shot planning from offline data without rewards or demonstrations.
-
GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulation
GR-2 pre-trains on web-scale videos then fine-tunes on robot data to reach 97.7% average success across over 100 manipulation tasks with strong generalization to new scenes and objects.
-
Gen2Act: Human Video Generation in Novel Scenarios enables Generalizable Robot Manipulation
Gen2Act enables generalizable robot manipulation for unseen objects and novel motions by using zero-shot human video generation from web data to condition a policy trained on an order of magnitude less robot interaction data.
-
Mobile ALOHA: Learning Bimanual Mobile Manipulation with Low-Cost Whole-Body Teleoperation
A low-cost whole-body teleoperation system enables effective imitation learning for complex bimanual mobile manipulation by co-training on mobile and static demonstration datasets.
-
Vision-Language Foundation Models as Effective Robot Imitators
RoboFlamingo adapts open-source vision-language models for robot manipulation tasks via single-step comprehension plus an explicit policy head, outperforming prior methods on benchmarks with only light fine-tuning.
-
Lift3D-VLA: Lifting VLA Models to 3D Geometry and Dynamics-Aware Manipulation
Lift3D-VLA integrates 3D point cloud encoding and temporal action modeling into Vision-Language-Action models, achieving higher success rates on simulated and real-world robotic manipulation tasks.
-
Not All Actions Are Equal: Rethinking Conditioning for Dexterous World Model
DexAC-WM improves FID, FVD, and PCK in high-DoF action-conditioned video prediction via structured action modeling and semantic grounding on EgoDex and EgoVerse.
-
World Value Models for Robotic Manipulation
World Value Model (WVM) integrates world models with value estimation to achieve SOTA Value-Order Correlation on expert and suboptimal robotic data and improves downstream policy performance.
-
Learning Dexterous Manipulation Using Contact Wrench Guidance From Human Demonstration
CHORD uses object-centric contact wrench guidance to improve RL scalability for long-horizon dexterous manipulation, reporting 82.12% average success on 1,831 of 4,739 benchmark tasks with real-world transfer.
-
Factor-Aware Mixture-of-Experts with Pretrained Encoder for Combinatorial Generalization
FAME combines a factor-aware MoE with frozen pretrained encoders via staged adapter training and joint fine-tuning, reporting 34% gains on Meta-World and 35% in real-world pick-and-place under environmental changes.
-
DexPIE: Stable Dexterous Policy Improvement from Real-World Experience
DexPIE improves dexterous manipulation success rates by 37% over demo policies via real-world experience collection with adapted intervention, multi-stage DAgger, asynchronous relative-action inference, and optimality conditioning.
-
ReGIL: Retrieval-Guided Imitation Learning from a Single Demonstration
ReGIL retrieves segments from a single demonstration to compute local temporal-alignment rewards and guide policy training, achieving >75% success on three real-robot tasks with <1 hour of online data.