Research map / Probabilistic and scientific modeling / Molecules and proteins
Implicit Latent Variable Model for Scene-Consistent Motion Forecasting
Sergio Casas, Cole Gulino, Simon Suo, Katie Luo, Renjie Liao, Raquel Urtasun
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 |
|---|---|---|---|
| use implicit latent output with deterministic decoder instead of explicit per-actor Gaussian decoder (ablation) | ILVM · ATG4D | SCR 5s (%): 8.46 -> 0.70 | Improved |
| use implicit decoder instead of explicit decoder (ablation) | ILVM · ATG4D | mean SADE (m): 1.80 -> 1.02 | Improved |
| use learned prior network instead of fixed isotropic Gaussian prior (ablation) | ILVM · ATG4D | SCR 5s (%): 1.10 -> 0.70 | Improved |
| use learned prior, mean SADE (ablation) | ILVM · ATG4D | mean SADE (m): 1.08 -> 1.02 | Improved |
| use graph-based SIM encoder and prior instead of actor-level MLPs (ablation) | ILVM · ATG4D | SCR 5s (%): 1.03 -> 0.70 | Improved |
| use SIM decoder instead of per-actor MLP decoder (ablation) | ILVM · ATG4D | SCR 5s (%): 1.52 -> 0.70 | Improved |
| use SIM encoder and decoder together instead of both MLPs (ablation) | ILVM · ATG4D | SCR 5s (%): 1.74 -> 0.70 | Improved |
| use SIM decoder, min SFDE (ablation) | ILVM · ATG4D | min SFDE (m): 1.67 -> 1.53 | Improved |
| replace baseline forecasting with ILVM in the planner | motion planner · ATG4D | planning collision rate up to 5s (%): 3.10 -> 2.64 | Improved |
| replace ESP forecasts with ILVM in the planner | motion planner · ATG4D | L2 to human at 5s (m): 5.48 -> 5.33 | Improved |
| use ILVM instead of best baseline on nuScenes | scene forecasting · nuScenes | SCR 5s (%): 4.56 -> 1.91 | Improved |
| share social features among actors without joint sampling | SpAGNN, SocialLSTM, NRI, MFP · ATG4D | SCR 5s (%) | No clear effect |
Closest papers by meaning
- SEPT: Towards Efficient Scene Representation Learning for Motion Prediction (ICLR 2024 · similarity 0.88)
- Motion Forecasting in Continuous Driving (NeurIPS 2024 · similarity 0.87)
- Learning Cooperative Trajectory Representations for Motion Forecasting (NeurIPS 2024 · similarity 0.87)
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.87)
- HPNet: Dynamic Trajectory Forecasting with Historical Prediction Attention (CVPR 2024 · similarity 0.86)
- SmartPretrain: Model-Agnostic and Dataset-Agnostic Representation Learning for Motion Prediction (ICLR 2025 · similarity 0.86)
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.
