Research map / Probabilistic and scientific modeling / Molecules and proteins
PRANK: motion Prediction based on RANKing
Yuriy Biktairov, Maxim Stebelev, Irina Rudenko, Oleh Shliazhko, Boris Yangel
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 trajectory ranking from a bank rather than an LSTM decoder | PRANK on in-house test · In-house test | ADE: 1.428 -> 1.302 | Improved |
| Use trajectory ranking from a bank rather than an LSTM decoder | PRANK on in-house test · In-house test | FDE: 3.152 -> 2.968 | Improved |
| Add trajectory noise (ablation) | PRANK · In-house test | FDE: 3.017 -> 2.968 | Improved |
| Add trajectory noise (ablation) | PRANK · In-house test | ADE: 1.302 -> 1.302 | No clear effect |
| Use scene history instead of only one timestamp (ablation) | PRANK · In-house test | ADE: 1.339 -> 1.302 | Improved |
| Add clustering-based rebalancing of h(t) (ablation) | PRANK · In-house test | ADE: 1.434 -> 1.302 | Improved |
| Add clustering-based rebalancing of h(t) (ablation) | PRANK · In-house test | LL: -2.62 -> -2.12 | Improved |
| Increase the mixture mode count (ablation) | PRANK multimodal · In-house test | LL: -2.12 -> -1.70 (2 modes), -1.61 (3 modes), -1.56 (5 modes) | Improved |
| Use the posterior mean rather than top-1 mode as the prediction | PRANK without noise · In-house test | ADE: 1.450 -> 1.302 | Improved |
| Increase posterior-mean search results from 150 to 1000 | PRANK without noise · In-house test | ADE: 1.302 -> 1.287 | Improved |
| Increase trajectory-bank index size | PRANK · In-house test | ADE: 1k: 2.3; 2M: 1.302; 500k: 1.298 | Mixed |
| Apply in-house trajectories to Argoverse | PRANK · Argoverse | ADE: 1.73 -> 1.84 | Hurt |
Closest papers by meaning
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.85)
- Motion Forecasting in Continuous Driving (NeurIPS 2024 · similarity 0.85)
- EDA: Evolving and Distinct Anchors for Multimodal Motion Prediction (AAAI 2024 · similarity 0.85)
- SmartPretrain: Model-Agnostic and Dataset-Agnostic Representation Learning for Motion Prediction (ICLR 2025 · similarity 0.85)
- SMART: Scalable Multi-agent Real-time Motion Generation via Next-token Prediction (NeurIPS 2024 · similarity 0.85)
- SmartRefine: A Scenario-Adaptive Refinement Framework for Efficient Motion Prediction (CVPR 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.
