Research map / Multi-agent systems and games / Human-AI and goal-conditioned
D2-TPred: Discontinuous Dependency for Trajectory Prediction under Traffic Lights
Yuzhen Zhang, Wentong Wang, Weizhi Guo, Pei Lv, Mingliang Xu, Wei Chen, Dinesh Manocha
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 SDG instead of GAT for spatial interaction (ablation) | D2-TPred with BDG · VTP-TL (with traffic lights) | ADE: 19.635 -> 16.900 | Improved |
| add SDG instead of GAT for spatial interaction (ablation) | D2-TPred with BDG · VTP-TL (with traffic lights) | FDE: 41.804 -> 34.553 | Improved |
| add BDG instead of LSTM for behavior dependency (ablation) | D2-TPred with SDG · VTP-TL (with traffic lights) | ADE: 20.082 -> 16.900 | Improved |
| add BDG instead of LSTM for behavior dependency (ablation) | D2-TPred with SDG · VTP-TL (with traffic lights) | FDE: 44.560 -> 34.553 | Improved |
| add SDG and BDG to GAT+LSTM baseline (ablation) | D2-TPred · VTP-TL (with traffic lights) | ADE: 21.792 -> 16.900 | Improved |
| add discriminator (ablation) | D2-TPred · VTP-TL (with traffic lights) | ADE: 18.626 -> 16.900 | Improved |
| add discriminator (ablation) | D2-TPred · VTP-TL (with traffic lights) | FDE: 39.598 -> 34.553 | Improved |
| encode traffic light states with MLP instead of LSTM (ablation) | D2-TPred · VTP-TL (with traffic lights) | ADE: 17.896 -> 16.900 | Improved |
| encode traffic light states with MLP instead of LSTM (ablation) | D2-TPred · VTP-TL (with traffic lights) | FDE: 37.629 -> 34.553 | Improved |
| add traffic light states as input (ablation) | D2-TPred · VTP-TL (with traffic lights) | ADE: 20.685 -> 16.900 | Improved |
| add traffic light states as input (ablation) | D2-TPred · VTP-TL (with traffic lights) | FDE: 47.296 -> 34.553 | Improved |
| add traffic light states as input (ablation) | Trajectron++ · VTP-TL (with traffic lights) | FDE: 118.37 -> 114.365 | No clear effect |
Closest papers by meaning
- Density-Adaptive Model Based on Motif Matrix for Multi-Agent Trajectory Prediction (CVPR 2024 · similarity 0.86)
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.84)
- C2F-TP: A Coarse-to-Fine Denoising Framework for Uncertainty-Aware Trajectory Prediction (AAAI 2025 · similarity 0.84)
- LaKD: Length-agnostic Knowledge Distillation for Trajectory Prediction with Any Length Observations (NeurIPS 2024 · similarity 0.84)
- Trajeglish: Traffic Modeling as Next-Token Prediction (ICLR 2024 · similarity 0.84)
- Towards Dynamic Spatial-Temporal Graph Learning: A Decoupled Perspective (AAAI 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.
