Research map / Multi-agent systems and games / Human-AI and goal-conditioned
MotionDiffuser: Controllable Multi-Agent Motion Prediction Using Diffusion
Chiyu “Max” Jiang, Andre Cornman, Cheolho Park, Benjamin Sapp, Yin Zhou, Dragomir Anguelov
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 Wayformer's prediction head with the MotionDiffuser diffusion head | Wayformer backbone · WOMD Interactive validation | minSADE: 0.99 -> 0.86 | Improved |
| replace Wayformer's prediction head with the MotionDiffuser diffusion head | Wayformer backbone · WOMD Interactive validation | mAP: 0.16 -> 0.19 | Improved |
| use MotionDiffuser instead of JFP | WOMD Interactive test · WOMD Interactive test | minSADE: 0.88 -> 0.86 | Improved |
| use MotionDiffuser instead of JFP | WOMD Interactive test · WOMD Interactive test | mAP: 0.21 -> 0.20 | Hurt |
| apply attractor guidance to target the ground-truth final point | MotionDiffuser · WOMD Interactive validation, attractor constraint | SR2m: 0.059 -> 0.952 | Improved |
| apply attractor guidance rather than post-sampling trajectory optimization | MotionDiffuser · WOMD Interactive validation, attractor constraint | minSADE: 4.563 -> 0.533 | Improved |
| apply repeller guidance between modeled agents | MotionDiffuser · WOMD Interactive validation, repeller constraint | Overlap: 0.059 -> 0.008 | Improved |
| add PCA-compressed trajectory representation (ablation) | MotionDiffuser · WOMD Interactive validation | minSADE: 1.03 -> 0.88 | Improved |
| use a Transformer denoiser instead of a simple MLP (ablation) | MotionDiffuser · WOMD Interactive validation | minSADE: 0.93 -> 0.88 | Improved |
| add self-attention between agents' noisy future trajectories (ablation) | MotionDiffuser · WOMD Interactive validation | minSADE: 0.91 -> 0.88 | Improved |
| add score thresholding to constraint guidance (ablation) | MotionDiffuser attractor sampling · WOMD Interactive validation, attractor constraint | minSFDE: 0.627 -> 0.007 | Improved |
Closest papers by meaning
- MotionLM: Multi-Agent Motion Forecasting as Language Modeling (ICCV 2023 · similarity 0.88)
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.87)
- Unified Uncertainty-Aware Diffusion for Multi-Agent Trajectory Modeling (CVPR 2025 · similarity 0.87)
- Multi-agent Long-term 3D Human Pose Forecasting via Interaction-aware Trajectory Conditioning (CVPR 2024 · similarity 0.86)
- Optimizing Diffusion Noise Can Serve As Universal Motion Priors (CVPR 2024 · similarity 0.86)
- Generative Trajectory Stitching through Diffusion Composition (NeurIPS 2025 · 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.
