Research map / Graph learning / Knowledge and temporal graphs
Learning Lane Graph Representations for Motion Forecasting
Ming Liang, Bin Yang, Rui Hu, Yun Chen, Renjie Liao, Song Feng, Raquel Urtasun
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 A2A (actor-to-actor) interaction to ActorNet (ablation) | ActorNet · Argoverse val (K=6) | minFDE: 1.66 -> 1.29 | Improved |
| add L2A to MapNet backbone (ablation) | ActorNet + MapNet · Argoverse val (K=6) | minFDE: 1.29 -> 1.23 | Improved |
| add A2L and L2L fusion to L2A (ablation) | MapNet + L2A · Argoverse val (K=6) | minFDE: 1.23 -> 1.10 | Improved |
| add A2A on top of L2A, A2L, L2L (ablation) | MapNet + L2A + A2L + L2L · Argoverse val (K=6) | minFDE: 1.10 -> 1.08 | Improved |
| add A2A on top of L2A, A2L, L2L (ablation) | MapNet + L2A + A2L + L2L · Argoverse val (K=1) | minFDE: 3.05 -> 2.97 | Improved |
| add multi-type connections to GraphConv (ablation) | GraphConv · Argoverse val (K=6) | minFDE: 1.41 -> 1.24 | Improved |
| add dilation to GraphConv (ablation) | GraphConv · Argoverse val (K=6) | minFDE: 1.41 -> 1.16 | Improved |
| add residual connection to GraphConv (ablation) | GraphConv · Argoverse val (K=6) | minFDE: 1.41 -> 1.33 | Improved |
| add dilation to residual multi-type LaneConv (ablation) | GraphConv + Residual + Multi-Type · Argoverse val (K=6) | minFDE: 1.19 -> 1.10 | Improved |
| add multi-type connections to residual dilated LaneConv (ablation) | GraphConv + Residual + Dilate · Argoverse val (K=6) | minFDE: 1.14 -> 1.10 | Improved |
| use LaneGCN model instead of leaderboard top entry Jean | Jean (1st) · Argoverse test (K=6) | minFDE: 1.49 -> 1.36 | Improved |
| use LaneGCN model instead of leaderboard top entry Jean | Jean (1st) · Argoverse test (K=1) | minFDE: 4.18 -> 3.78 | Improved |
Closest papers by meaning
- Learning and Aggregating Lane Graphs for Urban Automated Driving (CVPR 2023 · similarity 0.88)
- Generative Lane Topology Reasoning via Autoregressive Model with Geometry Prior (ECCV 2026 · similarity 0.88)
- LaneGraph2Seq: Lane Topology Extraction with Language Model via Vertex-Edge Encoding and Connectivity Enhancement (AAAI 2024 · similarity 0.86)
- LaneSegNet: Map Learning with Lane Segment Perception for Autonomous Driving (ICLR 2024 · similarity 0.86)
- SEPT: Towards Efficient Scene Representation Learning for Motion Prediction (ICLR 2024 · similarity 0.86)
- Learning Cooperative Trajectory Representations for Motion Forecasting (NeurIPS 2024 · similarity 0.85)
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.
