Research map / Multi-agent systems and games / Human-AI and goal-conditioned
SCAN: A Spatial Context Attentive Network for Joint Multi-Agent Intent Prediction
Jasmine Sekhon, Cody Fleming
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 interleaved spatial and temporal attention rather than omit decoder temporal attention (ablation) | SCAN · Average across five datasets | FDE (m): 1.04 -> 0.97 | Improved |
| Replace graph attention with SCAN's spatial attention (ablation) | SCAN · Average across five datasets | ADE (m): 0.63 -> 0.50 | Improved |
| Replace graph attention with SCAN's spatial attention (ablation) | SCAN · Average across five datasets | FDE (m): 1.27 -> 0.97 | Improved |
| Use a prediction horizon of 20 rather than 12 steps (ablation) | SCAN · Average across five datasets | ADE (m): 0.50 -> 0.98 | Hurt |
| Use a prediction horizon of 20 rather than 12 steps (ablation) | SCAN · Average across five datasets | FDE (m): 0.97 -> 2.04 | Hurt |
| Increase variety-loss sample count k (ablation) | generativeSCAN · Generated trajectory diversity | trajectory diversity | No clear effect |
| Add diversity loss with a larger weight lambda while using smaller k (ablation) | generativeSCAN · Generated trajectories in crossing scenario | trajectory diversity | Improved |
| Use a finer or coarser bearing and heading discretization than 30 degrees (ablation) | SCAN · Zara1 test | ADE | Hurt |
| Use SCAN rather than vanillaSCAN without decoder temporal attention (ablation) | SCAN · Average across five datasets | ADE (m): 0.52 -> 0.50 | Improved |
Closest papers by meaning
- Social-STGCNN: A Social Spatio-Temporal Graph Convolutional Neural Network for Human Trajectory Prediction (CVPR 2020 · similarity 0.88)
- Dynamic and Static Context-aware LSTM for Multi-agent Motion Prediction (ECCV 2020 · similarity 0.88)
- Social-Transmotion: Promptable Human Trajectory Prediction (ICLR 2024 · similarity 0.88)
- SocialMOIF: Multi-Order Intention Fusion for Pedestrian Trajectory Prediction (CVPR 2025 · similarity 0.88)
- Higher-order Relational Reasoning for Pedestrian Trajectory Prediction (CVPR 2024 · similarity 0.87)
- A Unified Environmental Network for Pedestrian Trajectory Prediction (AAAI 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.
