Data to 7 Oct 2026

Sparse Diffusion Policy: A Sparse, Reusable, and Flexible Policy for Robot Learning

arXiv 2407.01531 ↗First posted 1 Jul 2024cs.RO cs.LGcited by 56

On the map: Diffusion models → Diffusion and flow policies for robotics

also in 1 more

Mixture-of-experts → MoE in reinforcement learning and robotics

Abstract (arXiv)

The increasing complexity of tasks in robotics demands efficient strategies for multitask and continual learning. Traditional models typically rely on a universal policy for all tasks, facing challenges such as high computational costs and catastrophic forgetting when learning new tasks. To address these issues, we introduce a sparse, reusable, and flexible policy, Sparse Diffusion Policy (SDP). By adopting Mixture of Experts (MoE) within a transformer-based diffusion policy, SDP selectively activates experts and skills, enabling efficient and task-specific learning without retraining the entire model. SDP not only reduces the burden of active parameters but also facilitates the seamless integration and reuse of experts across various tasks. Extensive experiments on diverse tasks in both simulations and real world show that SDP 1) excels in multitask scenarios with negligible increases in active parameters, 2) prevents forgetting in continual learning of new tasks, and 3) enables efficient task transfer, offering a promising solution for advanced robotic applications. Demos and codes can be found in https://forrest-110.github.io/sparse_diffusion_policy/.

Key facts. Cited by 56 papers in the map (data through 7 Oct 2026). Placed in Diffusion and flow policies for robotics, one of 14 directions in Diffusion models. Builds on 70 of its 83 references that are in the map. First posted on arXiv 1 Jul 2024.

Builds on

70 of 83 references.

Closest ancestors · 7

Same direction, newest first

Other references · 63

2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
2016Overcoming catastrophic forgetting in neural networkscited by 4,718
2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,413
2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,014
2024Octo: An Open-Source Generalist Robot Policycited by 1,427
2016Progressive Neural Networkscited by 1,332
2019Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learningcited by 1,168
2018On First-Order Meta-Learning Algorithmscited by 1,003
2021GLaM: Efficient Scaling of Language Models with Mixture-of-Expertscited by 890
2023Open X-Embodiment: Robotic Learning Datasets and RT-X Modelscited by 849
2021Scaling Vision with Sparse Mixture of Expertscited by 740

38 more

2021What Matters in Learning from Offline Human Demonstrations for Robot Manipulationcited by 699
2022R3M: A Universal Visual Representation for Robot Manipulationcited by 678
2022Mixture-of-Experts with Expert Choice Routingcited by 611
2022Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulationcited by 545
2018Data-Efficient Hierarchical Reinforcement Learningcited by 436
2015Policy Distillationcited by 414
2022DeepSpeed-MoE: Advancing Mixture-of-Experts Inference and Training to Power Next-Generation AI Scalecited by 400
2023MimicGen: A Data Generation System for Scalable Robot Learning using Human Demonstrationscited by 308
2022Behavior Transformers: Cloning k modes with one stonecited by 299
2022Masked Visual Pre-training for Motor Controlcited by 246
2022Real-World Robot Learning with Masked Visual Pre-trainingcited by 241
2023RVT: Robotic View Transformer for 3D Object Manipulationcited by 224
2021Efficient Large Scale Language Modeling with Mixtures of Expertscited by 175
2015Conditional Computation in Neural Networks for faster modelscited by 166
2022On the Representation Collapse of Sparse Mixture of Expertscited by 156
2023Act3D: 3D Feature Field Transformers for Multi-Task Robotic Manipulationcited by 135
2023Distilled Feature Fields Enable Few-Shot Language-Guided Manipulationcited by 130
2020One Policy to Control Them All: Shared Modular Policies for Agent-Agnostic Controlcited by 114
2018Near-Optimal Representation Learning for Hierarchical Reinforcement Learningcited by 107
2024Uni-MoE: Scaling Unified Multimodal LLMs with Mixture of Expertscited by 107
2024Branch-Train-MiX: Mixing Expert LLMs into a Mixture-of-Experts LLMcited by 106
2017Vision-Based Multi-Task Manipulation for Inexpensive Robots Using End-To-End Learning from Demonstrationcited by 100
2023GNFactor: Multi-Task Real Robot Learning with Generalizable Neural Feature Fieldscited by 99
2022M3ViT: Mixture-of-Experts Vision Transformer for Efficient Multi-task Learning with Model-Accelerator Co-designcited by 97
2022Few-Shot Preference Learning for Human-in-the-Loop RLcited by 85
2023Bootstrap Your Own Skills: Learning to Solve New Tasks with Large Language Model Guidancecited by 85
2019Distilling Policy Distillationcited by 81
2023METRA: Scalable Unsupervised RL with Metric-Aware Abstractioncited by 80
2023DexArt: Benchmarking Generalizable Dexterous Manipulation with Articulated Objectscited by 64
2023LOTUS: Continual Imitation Learning for Robot Manipulation Through Unsupervised Skill Discoverycited by 63
2022Uni-Perceiver-MoE: Learning Sparse Generalist Models with Conditional MoEscited by 58
2021Hierarchical Reinforcement Learning By Discovering Intrinsic Optionscited by 57
2023Robot Learning with Sensorimotor Pre-trainingcited by 51
2023TAIL: Task-specific Adapters for Imitation Learning with Large Pretrained Modelscited by 48
2022PaCo: Parameter-Compositional Multi-Task Reinforcement Learningcited by 40
2022HetuMoE: An Efficient Trillion-scale Mixture-of-Expert Distributed Training Systemcited by 36
2019DisCoRL: Continual Reinforcement Learning via Policy Distillationcited by 35
2021Lifelong Robotic Reinforcement Learning by Retaining Experiencescited by 34

50 most cited of 63 All 63 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 6; 2025: 32; 2026: 18.

56 papers.

Sort:

By area:

2024TinyVLA: Towards Fast, Data-Efficient Vision-Language-Action Models for Robotic Manipulationcited by 331
2025DexVLA: Vision-Language Model with Plug-In Diffusion Expert for General Robot Controlcited by 191
2025ChatVLA: Unified Multimodal Understanding and Robot Control with Vision-Language-Action Modelcited by 112
2025Dita: Scaling Diffusion Transformer for Generalist Vision-Language-Action Policycited by 73
2024Diffusion-VLA: Generalizable and Interpretable Robot Foundation Model via Self-Generated Reasoningcited by 57
2025Token Reduction Should Go Beyond Efficiency in Generative Models -- From Vision, Language to Multimodalitycited by 49
2025ChatVLA-2: Vision-Language-Action Model with Open-World Embodied Reasoning from Pretrained Knowledgecited by 49
2024Efficient Diffusion Transformer Policies with Mixture of Expert Denoisers for Multitask Learningcited by 46
2025NavDP: Learning Sim-to-Real Navigation Diffusion Policy with Privileged Information Guidancecited by 46
2024Diffusion Transformer Policycited by 41
2025Failure Prediction at Runtime for Generative Robot Policiescited by 36
2024CARP: Visuomotor Policy Learning via Coarse-to-Fine Autoregressive Predictioncited by 26

38 more

50 most cited of 56 All 56 (JSON)

Similar, not cited

Close in wording, no citation either way