From Perception to Planning
Key papers behind image detection, 3D perception, motion prediction, and trajectory planning. Architecture explanations, comparisons, and tradeoffs.
Start with Article 1 · Full series guide
Browse all 12 articles
- Image perception: R-CNN, YOLO, and FCN
- Lidar perception: PointNet, VoxelNet, and CenterPoint
- Camera BEV: Lift-Splat-Shoot, BEVDepth, and BEVFormer
- Object queries: DETR3D, PETR, and Sparse4D
- Sensor fusion: PointPainting and BEVFusion
- Scene structure: tracking, MapTR, and road topology
- Occupancy and motion: SurroundOcc, Occ3D, and FIERY
- Motion prediction: VectorNet, LaneGCN, and MTR
- Trajectory planning: search, learned costs, and ChauffeurNet
- Joint perception and planning: TransFuser and UniAD
- Efficient and generative planning: VAD and DiffusionDrive
- Evaluating driving: nuPlan, PDM, NAVSIM, and world models