Research map / Probabilistic and scientific modeling / Molecules and proteins
Future-Aware Interaction Network For Motion Forecasting
Shijie Li, Chunyu Liu, Xun Xu, Si Yong Yeo, Xulei Yang
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 interaction-based design (future tokens in scene encoding) instead of MLP-based decoder (ablation) | FINet · Argoverse 2 (ablation) | minFDE6: 1.45 -> 1.27 | Improved |
| use interaction-based design instead of query-based decoder (ablation) | FINet · Argoverse 2 (ablation) | minFDE6: 1.43 -> 1.27 | Improved |
| use interaction-based design instead of MLP-based decoder (ablation) | FINet · Argoverse 2 (ablation) | MR6: 0.20 -> 0.15 | Improved |
| use interaction-based design instead of query-based decoder (ablation) | FINet · Argoverse 2 (ablation) | minFDE1: 4.28 -> 3.94 | Improved |
| add inductive bias on first future token (t=0 bias) (ablation) | FINet · Argoverse 2 (ablation) | minFDE6: 1.32 -> 1.27 | Improved |
| apply inductive bias on first token only rather than all future tokens (ablation) | FINet · Argoverse 2 (ablation) | minFDE6: 1.35 -> 1.27 | Improved |
| apply inductive bias to all tokens instead of none (ablation) | FINet · Argoverse 2 (ablation) | minFDE6: 1.32 -> 1.35 | Hurt |
| MLP-based decoder vs query-based decoder inside FINet (ablation) | FINet · Argoverse 2 (ablation) | minADE6: 0.74 -> 0.73 | No clear effect |
| replace QCNet transformer with Mamba-based FINet | FINet · Efficiency | latency (ms): 54.55 -> 17.72 | Improved |
| replace QCNet transformer with Mamba-based FINet | FINet · Efficiency | FLOPs (G): 28.0 -> 1.47 | Improved |
| replace QCNet transformer with Mamba-based FINet | FINet · Argoverse 2 test | minADE1: 1.69 -> 1.60 | Improved |
| replace QCNet transformer with Mamba-based FINet | FINet · Argoverse 2 test | minADE6: 0.65 -> 0.66 | No clear effect |
Closest papers by meaning
- Large Scale Interactive Motion Forecasting for Autonomous Driving: The Waymo Open Motion Dataset (ICCV 2021 · similarity 0.90)
- Trajectory Mamba: Efficient Attention-Mamba Forecasting Model Based on Selective SSM (CVPR 2025 · similarity 0.89)
- Motion Forecasting in Continuous Driving (NeurIPS 2024 · similarity 0.88)
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.88)
- HPNet: Dynamic Trajectory Forecasting with Historical Prediction Attention (CVPR 2024 · similarity 0.88)
- Future-Aware End-to-End Driving: Bidirectional Modeling of Trajectory Planning and Scene Evolution (NeurIPS 2025 · similarity 0.88)
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.
