Research map / Probabilistic and scientific modeling / Molecules and proteins
CoverNet: Multimodal Behavior Prediction Using Trajectory Sets
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Classify over a hybrid fixed-and-dynamic trajectory set instead of using regression baselines | CoverNet · nuScenes | HitRate: 0.13 -> 0.33 | Helped |
| Classify over a hybrid fixed-and-dynamic trajectory set instead of using MultiPath 64 | CoverNet · nuScenes | ADE15: 1.86 -> 1.28 | Helped |
| Use the hybrid trajectory set rather than the strongest listed baseline | CoverNet · nuScenes | FDE: 9.09 -> 9.26 | Hurt |
| Use CoverNet fixed ε=2 rather than MultiPath 64 | CoverNet · Internal dataset | HitRate: 0.27 -> 0.57 | Helped |
| Use CoverNet fixed ε=2 rather than MTP 1 | CoverNet · Internal dataset | ADE1: 1.88 -> 2.16 | Hurt |
| Use average rather than max point-wise distance for ground-truth matching (ablation) | CoverNet fixed trajectory set, 150 modes · Internal dataset, 3 sec horizon | ADE1: 1.0 -> 0.96 | Helped |
| Use average rather than max point-wise distance for ground-truth matching (ablation) | CoverNet fixed trajectory set, 150 modes · Internal dataset, 3 sec horizon | ADE5: 0.67 -> 0.66 | Helped |
| Use average rather than RMS point-wise distance for ground-truth matching (ablation) | CoverNet fixed trajectory set, 150 modes · Internal dataset, 3 sec horizon | ADE15: 0.63 -> 0.64 | Hurt |
| Add dynamic trajectories to a fixed trajectory set (ablation) | CoverNet trajectory-set generation · Trajectory-set coverage | number of trajectories needed for coverage | Helped |
| Increase fixed trajectory-set coverage by using ε=2 instead of ε=3 (ablation) | CoverNet fixed · nuScenes | HitRate: 0.23 -> 0.24 | Helped |
| Increase fixed trajectory-set coverage by using ε=2 instead of ε=3 (ablation) | CoverNet fixed · nuScenes | ADE15: 1.51 -> 1.63 | Hurt |
| Increase fixed trajectory-set coverage by using ε=2 instead of ε=3 (ablation) | CoverNet fixed · Internal dataset | HitRate: 0.33 -> 0.57 | Helped |
Closest papers by meaning
- BAT: Behavior-Aware Human-Like Trajectory Prediction for Autonomous Driving (AAAI 2024 · similarity 0.88)
- ModeSeq: Taming Sparse Multimodal Motion Prediction with Sequential Mode Modeling (CVPR 2025 · similarity 0.88)
- DriveGPT: Scaling Autoregressive Behavior Models for Driving (ICML 2025 · similarity 0.88)
- Trajeglish: Traffic Modeling as Next-Token Prediction (ICLR 2024 · similarity 0.87)
- Leveraging Driver Field-of-View for Multimodal Ego-Trajectory Prediction (ICLR 2025 · similarity 0.87)
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.87)
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.