Research map / Probabilistic and scientific modeling / Dynamical systems
Controlled SDEs for Long-Horizon Motion Generation under Latent Decision Uncertainty
Han Zhang, Nenggan Zheng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add command-modulated drift to MoFlow (M1) (ablation) | MoFlow · Rat Locomotion, horizon 60 | min ADE: 6.34 -> 4.12 | Improved |
| add stochastic diffusion to MoFlow (M2) (ablation) | MoFlow · Rat Locomotion, horizon 60 | min ADE: 6.34 -> 3.55 | Improved |
| add both drift modulation and diffusion to MoFlow (M3, full CogSDE) (ablation) | MoFlow · Rat Locomotion, horizon 60 | average ADE: 17.41 -> 10.41 | Improved |
| add diffusion on top of drift modulation (M1 to M3) (ablation) | CogSDE · Rat Locomotion, horizon 60 | min ADE: 4.12 -> 3.79 | Hurt |
| add drift modulation on top of diffusion (M2 to M3) (ablation) | CogSDE · Rat Locomotion, horizon 60 | min ADE: 3.55 -> 3.79 | Mixed |
| add drift modulation on top of diffusion (M2 to M3) (ablation) | CogSDE · Rat Locomotion, horizon 60 | average ADE: 12.60 -> 10.41 | Improved |
| add boundedness regularizer lambda=0.001 (ablation) | CogSDE · Rat Locomotion, horizon 60 | min ADE: 3.79 -> 3.65 | Improved |
| add boundedness regularizer lambda=0.001 (ablation) | CogSDE · Rat Locomotion, horizon 60 | average ADE: 10.42 -> 12.37 | Hurt |
| add boundedness regularizer lambda=0.003 (ablation) | CogSDE · Rat Locomotion, horizon 60 | average ADE: 10.42 -> 11.48 | Hurt |
| replace MoFlow with CogSDE | MoFlow · Rat Locomotion, horizon 60 | min ADE: 6.34 -> 3.79 | Improved |
| use CogSDE instead of command-aware SLD-HMP | SLD-HMP · BABEL, horizon 60 | min ADE: 0.023 -> 0.018 | Improved |
| use CogSDE instead of SLD-HMP on Rat Locomotion | SLD-HMP · Rat Locomotion, horizon 60 | min ADE: 3.89 -> 3.79 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MoFlow — MoFlow: One-Step Flow Matching for Human Trajectory Forecasting via Implicit Maximum Likelihood Estimation based Distillation (CVPR 2025) · better on 8 of 8 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Stable Long-Horizon PDE Forecasting via Latent Structured Spectral Propagators (NeurIPS 2026 · similarity 0.85)
- Simplifying Latent Dynamics with Softly State-Invariant World Models (NeurIPS 2024 · similarity 0.85)
- Reinforcement Learning Under Latent Dynamics: Toward Statistical and Algorithmic Modularity (NeurIPS 2024 · similarity 0.85)
- Closing the Approximation Gap in Simulation-free Latent SDEs (NeurIPS 2026 · similarity 0.84)
- Identifying latent state transitions in non-linear dynamical systems (ICLR 2025 · similarity 0.84)
- Efficient Dynamics Modeling in Interactive Environments with Koopman Theory (ICLR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
