LiDAR SLAM, HD Maps & Sensor Fusion for Autonomous Localization

Autonomous Vehicle Sensor Stack

  • GNSS RTK + INS: Absolute position anchor (Latitude, Longitude, Altitude, Roll, Pitch, Yaw). Fails inside urban canyons and tunnels.
  • 3D LiDAR (NDT / ICP Scan Matching): Aligns real-time point clouds against pre-mapped 3D point cloud layers of HD maps.
  • IMU & Wheel Odometry (Extended Kalman Filter / Factor Graph): Fills high-frequency gaps (100 Hz - 1000 Hz) between LiDAR/GNSS scans.

1. Why GNSS Alone Fails in Urban Canyons

Satellite multipath signals reflecting off tall glass skyscrapers introduce multi-meter errors into raw GNSS coordinates. To maintain lane-level accuracy (< 10 cm), autonomous driving software relies on Sensor Fusion combining HD vector maps with 3D LiDAR point cloud matching.