Skip to content

Perception

  • Sensor modalities: cameras (mono, stereo, fisheye), LiDAR (spinning, solid-state), radar, ultrasonic, IMU
  • Sensor fusion: early fusion (raw data), late fusion (decision level), multi-sensor calibration
  • 3D object detection: PointPillars, CenterPoint, BEVFusion, bird's-eye-view representations
  • Depth estimation: stereo matching, monocular depth networks, LiDAR-camera projection
  • Occupancy networks: 3D occupancy prediction, voxel representations
  • Lane detection and road topology: curve fitting, polynomial models, graph-based topology
  • Semantic mapping: building environmental representations from sensor streams