Research map / Probabilistic and scientific modeling / Molecules and proteins
Social-Mamba: Socially-Aware Trajectory Forecasting with State-Space Models
Po-Chien Luan, Wuyang Li, Yang Gao, Alexandre ALahi
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 |
|---|---|---|---|
| replace MHSA with Cycle Mamba as interaction block (ablation) | Social-Mamba · NBA-Full | ADE/FDE: 0.741/0.947 -> 0.719/0.919 | Improved |
| replace bidirectional Mamba with Cycle Mamba (ablation) | Social-Mamba · NBA-Full | ADE/FDE: 0.741/0.948 -> 0.719/0.919 | Improved |
| add ego module (with temporal and global, no goal) (ablation) | Social-Mamba · NBA-Full | ADE/FDE: 0.735/0.939 -> 0.729/0.928 | Improved |
| add goal module (with temporal and global, no ego) (ablation) | Social-Mamba · NBA-Full | ADE/FDE: 0.735/0.939 -> 0.727/0.928 | Improved |
| add both ego and goal modules (ablation) | Social-Mamba · NBA-Full | ADE/FDE: 0.735/0.939 -> 0.719/0.919 | Improved |
| run social triplet in parallel instead of sequential (ablation) | Social-Mamba · NBA-Full | ADE/FDE: 0.728/0.929 -> 0.719/0.919 | Improved |
| fuse social triplet with learnable weights instead of addition (ablation) | Social-Mamba · NBA-Full | ADE/FDE: 0.744/0.954 -> 0.719/0.919 | Improved |
| use Mamba decoder instead of MLP decoder (ablation) | Social-Mamba · NBA-Full | ADE/FDE: 0.730/0.931 -> 0.719/0.919 | Improved |
| Cycle Mamba vs MHSA efficiency (ablation) | Social-Mamba · NBA-Full | inference time (ms): 4.4 -> 3.4 | Improved |
| Cycle Mamba vs bidirectional Mamba efficiency (ablation) | Social-Mamba · NBA-Full | inference time (ms): 5.1 -> 3.4 | Improved |
| replace Transformer encoder in MoFlow with Social-Mamba | MoFlow · NBA-LED (MoFlow) | ADE/FDE: 0.71/0.87 -> 0.70/0.85 | Improved |
| Social-Mamba vs OmniTraj (pretrained) on NBA-Full | OmniTraj · NBA-Full | ADE/FDE: 0.73/0.94 -> 0.72/0.92 | Improved |
Closest papers by meaning
- Following the Human Thread in Social Navigation (ICLR 2025 · similarity 0.88)
- SocialMOIF: Multi-Order Intention Fusion for Pedestrian Trajectory Prediction (CVPR 2025 · similarity 0.88)
- Social-Transmotion: Promptable Human Trajectory Prediction (ICLR 2024 · similarity 0.87)
- SocialCircle: Learning the Angle-based Social Interaction Representation for Pedestrian Trajectory Prediction (CVPR 2024 · similarity 0.87)
- SocialCVAE: Predicting Pedestrian Trajectory via Interaction Conditioned Latents (AAAI 2024 · similarity 0.87)
- Trajectory Mamba: Efficient Attention-Mamba Forecasting Model Based on Selective SSM (CVPR 2025 · 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.
