Data to 7 Oct 2026

PoCo: Policy Composition from and for Heterogeneous Robot Learning

arXiv 2402.02511 ↗First posted 4 Feb 2024cs.RO cs.LGcited by 42

Not on the map. Why?

Cited by 20 in Diffusion models, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Training general robotic policies from heterogeneous data for different tasks is a significant challenge. Existing robotic datasets vary in different modalities such as color, depth, tactile, and proprioceptive information, and collected in different domains such as simulation, real robots, and human videos. Current methods usually collect and pool all data from one domain to train a single policy to handle such heterogeneity in tasks and domains, which is prohibitively expensive and difficult. In this work, we present a flexible approach, dubbed Policy Composition, to combine information across such diverse modalities and domains for learning scene-level and task-level generalized manipulation skills, by composing different data distributions represented with diffusion models. Our method can use task-level composition for multi-task manipulation and be composed with analytic cost functions to adapt policy behaviors at inference time. We train our method on simulation, human, and real robot data and evaluate in tool-use tasks. The composed policy achieves robust and dexterous performance under varying scenes and tasks and outperforms baselines from a single data source in both simulation and real-world experiments. See https://liruiw.github.io/policycomp for more details .

Key facts. Cited by 42 papers in the map (data through 7 Oct 2026). Builds on 66 of its 89 references that are in the map. First posted on arXiv 4 Feb 2024.

Builds on

66 of 89 references.

References · 66

2015Deep Residual Learning for Image Recognitioncited by 51,809
2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
2020Language Models are Few-Shot Learnerscited by 32,199
2020Denoising Diffusion Probabilistic Modelscited by 20,851
2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
2021Diffusion Models Beat GANs on Image Synthesiscited by 8,185
2023Segment Anythingcited by 8,158
2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
2022Classifier-Free Diffusion Guidancecited by 5,717
2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,223
2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,730

38 more

2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,007
2022Imagen Video: High Definition Video Generation with Diffusion Modelscited by 1,765
2022Make-A-Video: Text-to-Video Generation without Text-Video Datacited by 1,725
2021Ego4D: Around the World in 3,000 Hours of Egocentric Videocited by 1,530
2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,490
2019Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learningcited by 1,168
2022Planning with Diffusion for Flexible Behavior Synthesiscited by 1,023
2018On First-Order Meta-Learning Algorithmscited by 1,003
2019An Introduction to Variational Autoencoderscited by 997
2017An Overview of Multi-Task Learning in Deep Neural Networkscited by 791
2022R3M: A Universal Visual Representation for Robot Manipulationcited by 678
2022Compositional Visual Generation with Composable Diffusion Modelscited by 619
2017Sim-to-Real Transfer of Robotic Control with Dynamics Randomizationcited by 616
2017Generative Adversarial Networks: An Overviewcited by 588
2022Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulationcited by 545
2022BC-Z: Zero-Shot Task Generalization with Robotic Imitation Learningcited by 517
2022Is Conditional Generative Modeling all you need for Decision-Making?cited by 504
2018Reinforcement Learning and Control as Probabilistic Inference: Tutorial and Reviewcited by 466
2019Generalizing from a Few Examples: A Survey on Few-Shot Learningcited by 462
2021Sentence-T5: Scalable Sentence Encoders from Pre-trained Text-to-Text Modelscited by 462
2018An Algorithmic Perspective on Imitation Learningcited by 455
2022XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Modelcited by 436
2020Sim-to-Real Transfer in Deep Reinforcement Learning for Robotics: a Surveycited by 388
2019Which Tasks Should Be Learned Together in Multi-task Learning?cited by 367
2020Understanding Human Hands in Contact at Internet Scalecited by 281
2023Diffusion-based Generation, Optimization, and Planning in 3D Scenescited by 237
2017One-Shot Imitation Learningcited by 232
2023Goal-Conditioned Imitation Learning using Score-based Diffusion Policiescited by 229
2023Reduce, Reuse, Recycle: Compositional Generation with Energy-Based Diffusion Models and MCMCcited by 216
2023MimicPlay: Long-Horizon Imitation Learning by Watching Human Playcited by 214
2022SE(3)-DiffusionFields: Learning smooth cost functions for joint grasp and motion optimization through diffusioncited by 125
2023GenSim: Generating Robotic Simulation Tasks via Large Language Modelscited by 120
2022VideoDex: Learning Dexterity from Internet Videoscited by 101
2023Generative Skill Chaining: Long-Horizon Skill Planning with Diffusion Modelscited by 87
2019KETO: Learning Keypoint Representations for Tool Manipulationcited by 68
2021Controllable and Compositional Generation with Latent-Space Energy-Based Modelscited by 64
2019Rapid trial-and-error learning with simulation supports flexible tool use and physical reasoningcited by 60
2022ToolFlowNet: Robotic Manipulation with Tools via Predicting Tool Flow from Point Cloudscited by 54

50 most cited of 66 All 66 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 11; 2025: 23; 2026: 8.

42 papers.

Sort:

By area:

2024Diffusion Policy Policy Optimizationcited by 255
2025DexVLA: Vision-Language Model with Plug-In Diffusion Expert for General Robot Controlcited by 191
2025Steering Your Diffusion Policy with Latent Space Reinforcement Learningcited by 149
2024Reconciling Reality through Simulation: A Real-to-Sim-to-Real Approach for Robust Manipulationcited by 135
2025InstructVLA: Vision-Language-Action Instruction Tuning from Understanding to Manipulationcited by 71
2024Streaming Diffusion Policy: Fast Policy Synthesis with Variable Noise Diffusion Modelscited by 51
2024GenSim2: Scaling Robot Data Generation with Multi-modal and Reasoning LLMscited by 48
2024Transferable Tactile Transformers for Representation Learning Across Diverse Sensors and Taskscited by 44
2024STRAP: Robot Sub-Trajectory Retrieval for Augmented Policy Learningcited by 42
2025Diffusion Models for Robotic Manipulation: A Surveycited by 39
2025Generative Trajectory Stitching through Diffusion Compositioncited by 38
2024Deep Generative Models in Robotics: A Survey on Learning from Multimodal Demonstrationscited by 35

30 more

2024Inference-Time Policy Steering through Human Interactionscited by 35
2024DITTO: Demonstration Imitation by Trajectory Transformationcited by 28
2025Skill Expansion and Composition in Parameter Spacecited by 13
2025PolyTouch: A Robust Multi-Modal Tactile Sensor for Contact-rich Manipulation Using Tactile-Diffusion Policiescited by 13
2025A Survey on Imitation Learning for Contact-Rich Tasks in Roboticscited by 12
2025Multi-Modal Manipulation via Multi-Modal Policy Consensuscited by 9
2025CDP: Towards Robust Autoregressive Visuomotor Policy Learning via Causal Diffusioncited by 8
2025Uncertainty Comes for Free: Human-in-the-Loop Policies with Diffusion Modelscited by 6
2025Fine-tuning Diffusion Policies with Backpropagation Through Diffusion Timestepscited by 6
2025Projected Coupled Diffusion for Test-Time Constrained Joint Generationcited by 6
2026Multi-Camera View Scaling for Data-Efficient Robot Imitation Learningcited by 6
2026Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Trainingcited by 6
2025Modality-Composable Diffusion Policy via Inference-Time Distribution-level Compositioncited by 5
2025Learning to See and Act: Task-Aware Virtual View Exploration for Robotic Manipulationcited by 5
2025Flexible Multitask Learning with Factorized Diffusion Policycited by 5
2024Composing Diffusion Policies for Few-shot Learning of Movement Trajectoriescited by 3
2025The Unreasonable Effectiveness of Discrete-Time Gaussian Process Mixtures for Robot Policy Learningcited by 3
2025LodeStar: Long-horizon Dexterity via Synthetic Data Augmentation from Human Demonstrationscited by 3
2025How Well do Diffusion Policies Learn Kinematic Constraint Manifolds?cited by 3
2025Factorizing Diffusion Policies for Observation Modality Prioritizationcited by 2
2025MSG: Multi-Stream Generative Policies for Sample-Efficient Robotic Manipulationcited by 1
2026Meanings and Measurements: Multi-Agent Probabilistic Grounding for Vision-Language Navigationcited by 1
2026Factored Diffusion Policies:Compositionally Generalized Robot Control with a Single Score Networkcited by 1
2024ABNet: Attention BarrierNet for Safe and Scalable Robot Learningcited by 0
2025SeFA-Policy: Fast and Accurate Visuomotor Policy Learning with Selective Flow Alignmentcited by 0
2025Kinematics-Aware Diffusion Policy with Consistent 3D Observation and Action Space for Whole-Arm Robotic Manipulationcited by 0
2026Sparse Compositional Flow Matching by geometric assembly from motion primitivescited by 0
2026IMPACT: Learning Internal-Model Predictive Control for Forceful Robotic Manipulationcited by 0
2026Asynchronous Multimodal Diffusion Policy Composition via Latency-Aware Guidance Fusioncited by 0
2026Imperfection for Precision: Upcycling Imperfect Data for High-Precision Robotic Manipulationcited by 0

Similar, not cited

Close in wording, no citation either way