Research map / Probabilistic and scientific modeling / Dynamical systems
Latent-space control and planning
Plans and controls using learned latent dynamics, predictive coding and controlled SDEs.
78 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Dynamical systems, keyword cluster “physical · physics · simulation”
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.
14 papers in 2023–24 and 36 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 6.2% to 9.4%.
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
- 2016 · Learning to Poke by Poking: Experiential Learning of Intuitive Physics (NeurIPS 2016 · 275 citations)
- 2017 · Variational Laws of Visual Attention for Dynamic Scenes (NeurIPS 2017 · 24 citations)
- 2018 · Learning Plannable Representations with Causal InfoGAN (NeurIPS 2018 · 83 citations)
- 2019 · Learning-In-The-Loop Optimization: End-To-End Control And Co-Design Of Soft Robots Through Learned Deep Latent Representations (NeurIPS 2019 · 32 citations)
- 2020 · CoPhy: Counterfactual Learning of Physical Dynamics (ICLR 2020 · 34 citations)
- 2021 · Causal Navigation by Continuous-time Neural Networks (NeurIPS 2021 · 25 citations)
- 2022 · Filtered-CoPhy: Unsupervised Learning of Counterfactual Physics in Pixel Space (ICLR 2022 · 1 citation)
- 2023 · Improving Robotic Tactile Localization Super-resolution via Spatiotemporal Continuity Learning and Overlapping Air Chambers (AAAI 2023 · 9 citations)
- 2024 · Hybrid Internal Model: Learning Agile Legged Locomotion with Simulated Robot Response (ICLR 2024 · 4 citations)
- 2025 · Real-Time Execution of Action Chunking Flow Policies (NeurIPS 2025 · 7 citations)
- 2026 · Let EEG Models Learn EEG (ICML 2026 · 0 citations)
Earliest papers here
- Learning to Poke by Poking: Experiential Learning of Intuitive Physics (NeurIPS 2016 · 275 citations)
- Stability of Controllers for Gaussian Process Forward Models (ICML 2016 · 24 citations)
- A Bio-inspired Redundant Sensing Architecture (NeurIPS 2016)
What has been tried
50 papers here report 472 measured changes (a component added, removed or swapped, compared with a baseline): helped 66%, no clear effect 8%, hurt 22%, mixed 3%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Use asynchronous inpainting with soft masking instead of hard masking | Kinetix dynamic tasks · solve rate | Helped | Real-Time Execution of Action Chunking Flow Policies (NeurIPS 2025 · 7 citations) | |
| Add velocity inputs to the policy | Short-range stairs (A1) · success rate (%) | 85 -> 100 | Helped | Hybrid Internal Model: Learning Agile Legged Locomotion with Simulated Robot Response (ICLR 2024 · 4 citations) |
| increase the number of particles at inference time | MGL · position and angle recall | Mixed | Learning to be Smooth: An End-to-End Differentiable Particle Smoother (NeurIPS 2024 · 1 citation) | |
| Combine heterogeneous dynamics datasets in HDyS rather than train on one dataset | ImDy · mPJE | 0.6854 -> 0.5765 | Helped | Homogeneous Dynamics Space for Heterogeneous Humans (CVPR 2025 · 0 citations) |
| Use backward Euler instead of closed-form discretization | SC10, ρ = 10^-1 noise · accuracy | 79.3% vs 79.8% | No clear effect | UnHiPPO: Uncertainty-aware Initialization for State Space Models (ICML 2025 · 0 citations) |
| Use Euler integration instead of RK4 | Most Meta-World tasks · success rate | Helped | Time-Aware World Model for Adaptive Prediction and Control (ICML 2025 · 0 citations) | |
| add the measurement-induced bundle structure | Three experimental tasks (aggregate) · SR (%) | 62.7 -> 96.3 | Helped | Learning Dynamics under Environmental Constraints via Measurement-Induced Bundle Structures (ICML 2025 · 0 citations) |
| add cross-modal regularization to the forward-dynamics objective | All regimes and modalities · IQM | Helped | Dynamics-Aligned Latent Imagination in Contextual World Models for Zero-Shot Generalization (NeurIPS 2025 · 0 citations) | |
| Relax safety and reset thresholds from 0.1 to +∞ | 2D stochastic system · Safety MSE | 0.7010 ± 0.3847 -> 0.0023 ± 0.0065 | Helped | Safely Learning Controlled Stochastic Dynamics (NeurIPS 2025 · 0 citations) |
| Use sparse intermediate dimension 16 instead of 128 | Waymo speed control, relative speed changes ±50% · R² | 0.984 -> 0.770 | Hurt | Words in Motion: Extracting Interpretable Control Vectors for Motion Transformers (ICLR 2025 · 0 citations) |
| Remove distillation and use only iterative module selection | Meta-world · success rate | Hurt | Model Adaptation for Time Constrained Embodied Control (CVPR 2024 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 22 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Meta-World · 2 of 22 papers
- Push-T · 2 of 22 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 50 papers; from OpenAlex and the papers' affiliations).
- Massachusetts Institute of Technology · 4 papers
- Shanghai Jiao Tong University · 4 papers
- Tsinghua University · 2 papers
- Carnegie Mellon University · 2 papers
- Princeton University · 2 papers
- ETH Zurich · 2 papers
- University of California, Berkeley · 2 papers
- Microsoft · 2 papers
Use it
Download BibTeX of 39 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
- Physics-informed dynamics learning (127 papers)
- Latent dynamics models (111 papers)
- Data assimilation from sparse observations (58 papers)
- Stochastic population dynamics (53 papers)
Elsewhere in the topic
- State-space inference for stochastic systems (104 papers)
- Switching and law-discovery dynamics (81 papers)
- Synaptic plasticity learning rules (72 papers)
- Latent models of neural activity (70 papers)
- Recurrent circuit models of inference (70 papers)
- Low-rank and diffusion neural dynamics (53 papers)
- Functional connectivity modeling (50 papers)
- Stable contractive dynamics (42 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).