Logic
55 lines
9.0 KiB
JSON
55 lines
9.0 KiB
JSON
{
|
||
"candidates": [
|
||
{
|
||
"title": "Diffusion Policy",
|
||
"snippet": "# Diffusion Policy: Visuomotor Policy Learning via Action Diffusion\n[...]\nThis paper introduces Diffusion Policy, a new way of generating robot behavior by representing a robot’s visuomotor policy as a conditional denoising diffusion process. We benchmark Diffusion Policy across 12 different tasks from 4 different robot manipulation benchmarks and find that it consistently outperforms existing state-of-the-art robot learning methods with an average improvement of 46.9%. Diffusion Policy learns the gradient of the action-distribution score function and iteratively optimizes with respect to this gradient field during inference via a series of stochastic Langevin dynamics steps. We find that the diffusion formulation yields powerful advantages when used for robot policies, including gracefully handling multimodal action distributions, being suitable for high-dimensional action spaces, and exhibiting impressive training stability. To fully unlock the potential of diffusion models for visuomotor policy learning on physical robots, this paper presents a set of key technical contributions including the incorporation of receding horizon control, visual conditioning, and the time-series diffusion transformer. We hope this work will help motivate a new generation of policy learning techniques that are able to leverage the powerful generative modeling capabilities of diffusion models. Code, data, and training details will be publicly available.\n[...]\n[width=0.95]figure/DP_teaser.pdf \\ca",
|
||
"source_url": "https://arxiv.org/html/2502.12371v1",
|
||
"discovered_for": [
|
||
"related_work.diffusion_policy"
|
||
],
|
||
"_exa_id": "https://arxiv.org/html/2502.12371v1",
|
||
"_exa_published_date": "2025-01-01T00:00:00.000Z"
|
||
},
|
||
{
|
||
"title": "Visuomotor policy learning via action diffusion - ACM Digital Library",
|
||
"snippet": "Diffusion policy: : Visuomotor policy learning via action diffusion: International Journal of Robotics Research: Vol 44, No 10-11 other-periodical;requestedJournal:\n[...]\n:rbrs;wgroup:string:ACM Publication Websites;subPage:string:Basic Abstract;article:article:doi\\:10.\n[...]\n783649241273668;page:string:Article/Chapter View;ctype:string:Journal Content;groupTopic:topic:acm-pubtype>other-periodical;website:website:dl-site;issue:issue:doi\\:10.5555/rbrs.2025.44.issue-10-11;csubtype:string:Periodical;taxonomy:taxonomy:acm-pubtype;pageGroup:string:Publication Pages\"> skip to main content\n\n \n\n \n\nContents\n[...]\nThis paper introduces Diffusion Policy, a new way of generating robot behavior by representing a robot’s visuomotor policy as a conditional denoising diffusion process. We benchmark Diffusion Policy across 15 different tasks from 4 different robot manipulation benchmarks and find that it consistently outperforms existing state-of-the-art robot learning methods with an average improvement of 46.9%. Diffusion Policy learns the gradient of the action-distribution score function and iteratively optimizes with respect to this gradient field during inference via a series of stochastic Langevin dynamics steps. We find that the diffusion formulation yields powerful advantages when used for robot policies, including gracefully handling multimodal action distributions, being suitable for high-dimensional action spaces, and exhibiting impressive training stability. To fully unlock the p",
|
||
"source_url": "https://dl.acm.org/doi/10.1177/02783649241273668",
|
||
"discovered_for": [
|
||
"related_work.diffusion_policy"
|
||
],
|
||
"_exa_id": "https://dl.acm.org/doi/10.1177/02783649241273668",
|
||
"_exa_published_date": null
|
||
},
|
||
{
|
||
"title": "",
|
||
"snippet": "This paper introduces Diffusion Policy, a new way of generating robot behavior by representing a robot’s visuomotor\n[...]\npolicy as a conditional denoising diffusion process. We benchmark Diffusion Policy across 15 different tasks from 4\n[...]\ndifferent robot manipulation benchmarks and find that it consistently outperforms existing state-of-the-art robot learning\n[...]\nmethods with an average improvement of 46.9%. Diffusion Policy learns the gradient of the action-distribution score\n[...]\nfunction and iteratively optimizes with respect to this gradient field during inference via a series of stochastic Langevin\n[...]\ndynamics steps. We find that the diffusion formulation yields powerful advantages when used for robot policies, including\n[...]\ngracefully handling multimodal action distributions, being suitable for high-dimensional action spaces, and exhibiting\n[...]\nimpressive training stability. To fully unlock the potential of diffusion models for visuomotor policy learning on physical\n[...]\nrobots, this paper presents a set of key technical contributions including the incorporation of receding horizon control,\n[...]\nvisual conditioning, and the time-series diffusion transformer. We hope this work will help motivate a new generation of\n[...]\npolicy learning techniques that are able to leverage the powerful generative modeling capabilities of diffusion models.\n[...]\nintroducing a new form of robot visuomotor policy that\n[...]\ngenerates behavior via a “conditional denoising di",
|
||
"source_url": "https://arxiv.org/pdf/2303.04137v5",
|
||
"discovered_for": [
|
||
"related_work.diffusion_policy"
|
||
],
|
||
"_exa_id": "https://arxiv.org/pdf/2303.04137v5",
|
||
"_exa_published_date": null
|
||
},
|
||
{
|
||
"title": "Visuomotor Policy Learning via Action Diffusion - arXiv",
|
||
"snippet": "This paper introduces Diffusion Policy, a new way of generating robot behavior by representing a robot’s visuomotor policy as a conditional denoising diffusion process. We benchmark Diffusion Policy across 15 different tasks from 4 different robot manipulation benchmarks and find that it consistently outperforms existing state-of-the-art robot learning methods with an average improvement of 46.9%. Diffusion Policy learns the gradient of the action-distribution score function and iteratively optimizes with respect to this gradient field during inference via a series of stochastic Langevin dynamics steps. We find that the diffusion formulation yields powerful advantages when used for robot policies, including gracefully handling multimodal action distributions, being suitable for high-dimensional action spaces, and exhibiting impressive training stability. To fully unlock the potential of diffusion models for visuomotor policy learning on physical robots, this paper presents a set of key technical contributions including the incorporation of receding horizon control, visual conditioning, and the time-series diffusion transformer. We hope this work will help motivate a new generation of policy learning techniques that are able to leverage the powerful generative modeling capabilities of diffusion models. Code, data, and training details is available diffusion-policy.cs.columbia.edu\n[...]\nIn this work, we seek to address this challenge by introducing a new form of robot visuomoto",
|
||
"source_url": "https://arxiv.org/abs/2303.04137",
|
||
"discovered_for": [
|
||
"related_work.diffusion_policy"
|
||
],
|
||
"_exa_id": "https://arxiv.org/abs/2303.04137",
|
||
"_exa_published_date": "2023-03-07T00:00:00.000Z"
|
||
},
|
||
{
|
||
"title": "3D Diffusion Policy: Generalizable Visuomotor Policy Learning via Simple 3D Representations",
|
||
"snippet": "Imitation learning provides an efficient way to teach robots dexterous skills; however, learning complex skills robustly and generalizablely usually consumes large amounts of human demonstrations. To tackle this challenging problem, we present 3D Diffusion Policy (DP3), a novel visual imitation learning approach that incorporates the power of 3D visual representations into diffusion policies, a class of conditional action generative models. The core design of DP3 is the utilization of a compact 3D visual representation, extracted from sparse point clouds with an efficient point encoder. In our experiments involving 72 simulation tasks, DP3 successfully handles most tasks with just 10 demonstrations and surpasses baselines with a 24.2% relative improvement. In 4 real robot tasks, DP3 demonstrates precise control with a high success rate of 85%, given only 40 demonstrations of each task, and shows excellent generalization abilities in diverse aspects, including space, viewpoint, appearance, and instance. Interestingly, in real robot experiments, DP3 rarely violates safety requirements, in contrast to baseline methods which frequently do, necessitating human intervention. Our extensive evaluation highlights the critical importance of 3D representations in real-world robot learning. Videos, code, and data are available on 3d-diffusion-policy.github.io.\n[...]\nTo tackle this challenging problem, we introduce 3D Diffusion Policy (DP3), a simple yet effective visual imitation learnin",
|
||
"source_url": "https://arxiv.org/html/2403.03954v2",
|
||
"discovered_for": [
|
||
"related_work.diffusion_policy"
|
||
],
|
||
"_exa_id": "https://arxiv.org/html/2403.03954v2",
|
||
"_exa_published_date": null
|
||
}
|
||
]
|
||
}
|