Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Aware of the History: Trajectory Forecasting with the Local Behavior Data
Yiqi Zhong, Zhenyang Ni, Siheng Chen, Ulrich Neumann
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 local behavior data as input (LBA) | LaneGCN · nuScenes test | minADE1: 6.17 -> 2.72 | Improved |
| add local behavior data as input (LBA) | P2T · nuScenes test | minFDE1: 10.80 -> 9.25 | Improved |
| add local behavior data as input (LBA) | LaneGCN · Argoverse test | minFDE1: 3.89 -> 3.61 | Improved |
| add local behavior data as input (LBA) | DenseTNT · Argoverse test | minADE1: 1.70 -> 1.65 | Improved |
| add distilled local-behavior estimation (LBF) | LaneGCN · Argoverse test | minFDE1: 3.89 -> 3.54 | Improved |
| add distilled local-behavior estimation (LBF) | LaneGCN · nuScenes test | minADE1: 6.17 -> 5.58 | Improved |
| increase local range from 0.5m to 1.5m (ablation) | LaneGCN LBA · Argoverse test | minADE1: 1.64 -> 1.71 | Hurt |
| increase local range from 0.5m to 1.5m (ablation) | LaneGCN LBF · Argoverse test | minADE1: 1.61 -> 1.63 | No clear effect |
| add knowledge distillation loss (ablation) | LaneGCN LBF · Argoverse test | minADE1: 1.68 -> 1.61 | Improved |
| add LBA teacher instead of self-KD (ablation) | LaneGCN LBF · Argoverse test | minFDE1: 3.67 -> 3.63 | Improved |
| add behavior estimator (ablation) | LaneGCN LBF · Argoverse test | minFDE1: 3.63 -> 3.54 | Improved |
| use local behavior data for agents with very few trajectories (ablation) | LaneGCN LBA · Argoverse val | minFDE1: 3.33 -> 3.36 | Hurt |
Closest papers by meaning
- Modality Exploration, Retrieval and Adaptation for Trajectory Prediction (TPAMI 2023 · similarity 0.87)
- HPNet: Dynamic Trajectory Forecasting with Historical Prediction Attention (CVPR 2024 · similarity 0.87)
- STraj: Self-training for Bridging the Cross-Geography Gap in Trajectory Prediction (AAAI 2025 · similarity 0.86)
- MOVEBENCH: A Benchmark for Global-Scale Wildlife Movement Forecasting (NeurIPS 2026 · similarity 0.85)
- TrajAgent: An LLM-Agent Framework for Trajectory Modeling via Large-and-Small Model Collaboration (NeurIPS 2025 · similarity 0.85)
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.85)
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.
