Research map / Multi-agent systems and games / Human-AI and goal-conditioned
VisionTrap: Vision-Augmented Trajectory Prediction Guided by Textual Descriptions
Seokha Moon, Hyun Woo, Hongbeen Park, Haeji Jung, Reza Mahjourian, Hyung-gun Chi, Hyerin Lim, Sangpil Kim, Jinkyu Kim
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 Visual Semantic Encoder using surround-view images (ablation) | VisionTrap · nuScenes (whole dataset) | ADE_10: 0.407 -> 0.382 | Improved |
| add Text-driven Guidance with VLM/LLM descriptions (ablation) | VisionTrap · nuScenes (whole dataset) | ADE_10: 0.382 -> 0.368 | Improved |
| add Map Encoder (ablation) | VisionTrap baseline · nuScenes (whole dataset) | ADE_10: 0.425 -> 0.407 | Improved |
| add Visual Semantic Encoder (ablation) | VisionTrap · nuScenes (prediction split) | MR_10: 0.53 -> 0.36 | Improved |
| add visual and text modules (ablation) | VisionTrap · nuScenes (prediction split) | inference time (msec): 21 -> 53 | Hurt |
| use asymmetric contrastive loss instead of symmetric CLIP loss (ablation) | Text-driven Guidance · nuScenes (whole dataset) | ADE_6: 0.51 -> 0.44 | Improved |
| refine negative pairs to remove false negatives (ablation) | Text-driven Guidance · nuScenes (whole dataset) | ADE_6: 0.49 -> 0.44 | Improved |
| add top-k limit on number of negative pairs (ablation) | Text-driven Guidance · nuScenes (whole dataset) | ADE_6: 0.46 -> 0.44 | Improved |
| add vision and text to trajectory-only model for pedestrians (ablation) | VisionTrap · nuScenes (whole dataset, pedestrians) | MR_10: 0.019 -> 0.016 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- AgentFormer — AgentFormer: Agent-Aware Transformers for Socio-Temporal Multi-Agent Forecasting (ICCV 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- AutoTraces: Autoregressive Trajectory Forecasting via Multimodal Large Language Models (CVPR 2026 · similarity 0.88)
- Less Language, More Latents: Annotation-Efficient VLAs for Driving (NeurIPS 2026 · similarity 0.87)
- SOLVE: Synergy of Language-Vision and End-to-End Networks for Autonomous Driving (CVPR 2025 · similarity 0.87)
- Multi-modal Knowledge Distillation-based Human Trajectory Forecasting (CVPR 2025 · similarity 0.87)
- VLP: Vision Language Planning for Autonomous Driving (CVPR 2024 · similarity 0.86)
- SpaceDrive: Infusing Spatial Awareness into VLM-based Autonomous Driving (CVPR 2026 · 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.
