Research map / Image generation / Autoregressive and latent generation
Diffusion planning and policies
Use diffusion models for robot planning, behavior synthesis, and visuomotor control.
58 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2022–2026) · an approach in Autoregressive and latent generation, keyword cluster “motion · layout · object”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
18 papers in 2023–24 and 39 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 3.8% to 4.3%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2022 · Generating Useful Accident-Prone Driving Scenarios via a Learned Traffic Prior (CVPR 2022 · 143 citations)
- 2023 · Scenario Diffusion: Controllable Driving Scenario Generation With Diffusion (NeurIPS 2023 · 13 citations)
- 2024 · SingularTrajectory: Universal Trajectory Predictor Using Diffusion Model (CVPR 2024 · 67 citations)
- 2025 · DiffScene: Diffusion-Based Safety-Critical Scenario Generation for Autonomous Vehicles (AAAI 2025 · 42 citations)
- 2026 · From Noise to Intent: Anchoring Generative VLA Policies with Residual Bridges (ICML 2026 · 0 citations)
Earliest papers here
- Generating Useful Accident-Prone Driving Scenarios via a Learned Traffic Prior (CVPR 2022 · 143 citations)
- Scenario Diffusion: Controllable Driving Scenario Generation With Diffusion (NeurIPS 2023 · 13 citations)
- AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners (ICML 2023 · 7 citations)
What has been tried
38 papers here report 420 measured changes (a component added, removed or swapped, compared with a baseline): helped 70%, no clear effect 5%, hurt 20%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Use K=4 rather than K=1 motion vectors | ZARA1 all tasks average · ADE/FDE | 0.40 / 0.77 -> 0.30 / 0.59 | Helped | SingularTrajectory: Universal Trajectory Predictor Using Diffusion Model (CVPR 2024 · 67 citations) |
| Increase the number of adversarial optimization steps during each denoising step | Safety-critical scenario generation · Collision Rate | Helped | DiffScene: Diffusion-Based Safety-Critical Scenario Generation for Autonomous Vehicles (AAAI 2025 · 42 citations) | |
| add a discrete skill set to language-conditioned diffusion planning | Meta-World MT10 · performance | 16.7% -> 23.3% | Helped | SkillDiffuser: Interpretable Hierarchical Planning via Skill Abstractions in Diffusion-Based Task Execution (CVPR 2024 · 26 citations) |
| Add surface pulling force (SRF) | DexGraspNet test set · Suc.6 | 26.6 -> 38.9 | Helped | DexGrasp Anything: Towards Universal Robotic Dexterous Grasping with Physics Awareness (CVPR 2025 · 20 citations) |
| remove lane ordering | nuPlan lane graph generation · FD | 1.05 -> 1.36 | Hurt | Scenario Dreamer: Vectorized Latent Diffusion for Generating Driving Simulation Environments (CVPR 2025 · 11 citations) |
| Use closed-loop feedback for adaptive sub-goal selection instead of open-loop execution | CALVIN ABC→D · Avg. Len. | +0.44 average completed task length | Helped | Closed-Loop Visuomotor Control with Generative Expectation for Robotic Manipulation (NeurIPS 2024 · 10 citations) |
| Apply repeated projections during reverse diffusion instead of a single post-processing projection | Constrained Trajectories · feasibility / success percentage | Helped | Constrained Synthesis with Projected Diffusion Models (NeurIPS 2024 · 7 citations) | |
| Remove causal sparsity constraints from the scene encoder | nuScenes, multi-agent, K=5 · SCR | Hurt | Causal Composition Diffusion Model for Closed-loop Traffic Generation (CVPR 2025 · 6 citations) | |
| Add the dynamic affordance trajectory to contact-point-only conditioning | Simulation, unseen instance · success rate | 42.5% -> 50.0% | Helped | AffordDP: Generalizable Diffusion Policy with Transferable Affordance (CVPR 2025 · 5 citations) |
| Use global rather than local representation | Unconditioned · FID | 0.431 -> 0.161 | Helped | HandDiffuse: Generative Controllers for Two-Hand Interactions via Diffusion Models (AAAI 2025 · 4 citations) |
| add verbal and non-verbal motion memory banks | DyConv test set · SyncScore | 6.103 -> 7.188 | Helped | INFP: Audio-Driven Interactive Head Generation in Dyadic Conversations (CVPR 2025 · 4 citations) |
| Add random rotation augmentation during GraspVAE training | Ablation evaluation · IV | 4.17 -> 3.78 | Helped | LatentHOI: On the Generalizable Hand Object Motion Generation with Latent Hand Diffusion. (CVPR 2025 · 4 citations) |
What reviewers will expect
Datasets and baselines named in the 26 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Waymo Open Motion Dataset · 2 of 26 papers
- nuPlan · 2 of 26 papers
Baselines
- IDM · 3 of 26 papers
- MDM · 2 of 26 papers
- Diffusion Policy · 2 of 26 papers
- STRIVE · 2 of 26 papers
- UGG · 2 of 26 papers
- PDM-Closed · 2 of 26 papers
- Diffusion-ES · 2 of 26 papers
- PlanTF · 2 of 26 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 57 papers; from OpenAlex and the papers' affiliations).
- University of California, Berkeley · 4 papers
- Tsinghua University · 4 papers
- Massachusetts Institute of Technology · 3 papers
- The University of Hong Kong · 3 papers
- University of Virginia · 3 papers
- Carnegie Mellon University · 2 papers
- ShanghaiTech University · 2 papers
- Meta · 2 papers
- Harvard University · 2 papers
- Alibaba Group · 2 papers
Use it
Download BibTeX of 24 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Compositional scene generation (105 papers)
- Layout generation (39 papers)
Elsewhere in the topic
- Classical generative image synthesis (147 papers)
- High-resolution diffusion synthesis (140 papers)
- Retrieval and matching with generative features (140 papers)
- Generative latent space tooling (129 papers)
- Generalizable fake image detection (119 papers)
- Latent diffusion with learned tokenizers (114 papers)
- Synthetic training data from generators (111 papers)
- Next-scale and token autoregressive images (90 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).