Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Query-Centric Trajectory Prediction
Read the paperCheck 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 |
|---|---|---|---|
| Use QCNet with ensembling rather than without ensembling (ablation) | QCNet · Argoverse 2 test | minADE6: 0.65 -> 0.62 | Helped |
| Add factorized scene-context fusion blocks (ablation) | QCNet · Argoverse 2 validation | minADE6: 0.76 -> 0.74 -> 0.73 | Helped |
| Reuse cached scene encodings during online inference (ablation) | QCNet · Argoverse 2 validation, densest traffic scene | Online inference latency: 82±1 -> 13±1 ms | Helped |
| Add recurrence to the trajectory proposal module (ablation) | QCNet · Argoverse 2, 6-sec prediction | minFDE6: 1.47 -> 1.40 | Helped |
| Add the refinement module after recurrent proposals (ablation) | QCNet · Argoverse 2, 6-sec prediction | minFDE6: 1.40 -> 1.27 | Helped |
| Increase recurrent steps from 3 to 6 (ablation) | QCNet with refinement · Argoverse 2, 6-sec prediction | minFDE6: 1.27 -> 1.27 | No clear effect |
| Add recurrence to the proposal module (ablation) | QCNet without refinement · Argoverse 1, 3-sec prediction | minFDE6: 0.92 -> 0.90 -> 0.90 | No clear effect |
| Remove map and social context from the anchor-free proposal decoder (ablation) | QCNet proposal module · Argoverse 2 validation | minADE6 | Hurt |
| Remove Mode2Mode self-attention (ablation) | QCNet proposal or refinement module · Argoverse 2 validation | Long-term accuracy | Hurt |
| Use the full QCNet model instead of competing methods | QCNet · Argoverse 2 test | minFDE6: 1.44 -> 1.19 | Helped |
| Use the full QCNet model instead of the strongest listed baseline | QCNet · Argoverse 1 leaderboard | minFDE6: 1.14 -> 1.07 | Helped |
Closest papers by meaning
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.88)
- Motion Forecasting in Continuous Driving (NeurIPS 2024 · similarity 0.87)
- Trajectory Mamba: Efficient Attention-Mamba Forecasting Model Based on Selective SSM (CVPR 2025 · similarity 0.87)
- SmartRefine: A Scenario-Adaptive Refinement Framework for Efficient Motion Prediction (CVPR 2024 · similarity 0.87)
- AutoTraces: Autoregressive Trajectory Forecasting via Multimodal Large Language Models (CVPR 2026 · similarity 0.87)
- HPNet: Dynamic Trajectory Forecasting with Historical Prediction Attention (CVPR 2024 · similarity 0.86)
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.