Research map / Multi-agent systems and games / Human-AI and goal-conditioned
LaPred: Lane-Aware Prediction of Multi-Modal Future Trajectories of Dynamic Agents
Check 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 |
|---|---|---|---|
| add lane information with hard feature selection (ablation) | CNN-LSTM target-history baseline · nuScenes validation | ADE_1: 4.77 -> 4.04 | Helped |
| add lane-off loss (ablation) | LaPred with lane information and hard selection · nuScenes validation | ADE_5: 2.06 -> 1.77 | Helped |
| add nearby-agent histories (ablation) | LaPred with lane information and lane-off loss · nuScenes validation | ADE_5: 1.77 -> 1.72 | Helped |
| use soft weighted feature aggregation instead of hard selection (ablation) | LaPred with lanes and nearby agents · nuScenes validation | ADE_5: 1.72 -> 1.53 | Helped |
| use one selected nearby agent per lane rather than max-pooling multiple agents by closest lane (ablation) | LaPred · nuScenes validation | ADE_5: 1.56 -> 1.53 | Helped |
| use one selected nearby agent per lane rather than aggregating every agent to every lane (ablation) | LaPred · nuScenes validation | ADE_5: 1.60 -> 1.53 | Helped |
| use LaPred rather than MHA-JAM | trajectory prediction · nuScenes validation | FDE_10: 2.21 -> 2.39 | Hurt |
| use LaPred rather than MHA-JAM | trajectory prediction · nuScenes validation | ADE_1: 3.69 -> 3.51 | Helped |
| use LaPred rather than MTPLA | trajectory prediction · Argoverse validation | ADE_6: 1.05 -> 0.71 | Helped |
| use LaPred rather than MTPLA | trajectory prediction · Argoverse validation | ADE_1: 1.46 -> 1.48 | Hurt |
Closest papers by meaning
- Divide-and-Conquer for Lane-Aware Diverse Trajectory Prediction (CVPR 2021 · similarity 0.91)
- LaneSegNet: Map Learning with Lane Segment Perception for Autonomous Driving (ICLR 2024 · similarity 0.87)
- HPNet: Dynamic Trajectory Forecasting with Historical Prediction Attention (CVPR 2024 · similarity 0.87)
- Density-Adaptive Model Based on Motif Matrix for Multi-Agent Trajectory Prediction (CVPR 2024 · similarity 0.86)
- LaKD: Length-agnostic Knowledge Distillation for Trajectory Prediction with Any Length Observations (NeurIPS 2024 · similarity 0.86)
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 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.