Overview
Autonomous navigation research at NSL targets the demanding requirements of self-driving vehicles and unmanned aerial vehicles (UAVs). These applications require sub-decimeter positioning accuracy, continuous availability, and extremely high integrity — far beyond what conventional GNSS can provide alone. We develop multi-sensor fusion systems that meet these stringent requirements.
Research Topics
Autonomous Vehicle Localization
Self-driving cars require lane-level positioning accuracy (< 0.1 m) in all weather and lighting conditions. Our research integrates:
- GNSS/INS as the primary navigation backbone
- LiDAR point cloud matching against HD maps
- Camera-based lane detection and landmark recognition
- HD Map fusion for absolute position correction
- Vehicle odometry for additional constraints
UAV Navigation
Unmanned aerial vehicles face unique navigation challenges including high dynamics, weight constraints, and operation in both urban and GPS-degraded environments. We research:
- Lightweight GNSS/IMU systems for small UAVs
- Vision-aided navigation using optical flow and visual odometry
- Safe landing zone detection using onboard sensors
- Geofencing and corridor navigation using precise positioning
Simultaneous Localization and Mapping (SLAM)
When HD maps are unavailable, SLAM algorithms can simultaneously build a map while localizing within it. We develop:
- LiDAR-SLAM for outdoor environments
- Visual-inertial SLAM for indoor/GPS-denied scenarios
- GNSS-aided SLAM for globally consistent maps
- Tightly-coupled optimization using factor graphs
Deep Learning for Navigation
Machine learning is transforming navigation systems. Our research applies:
- Neural networks for GNSS signal quality assessment
- Deep learning-based pedestrian dead reckoning
- End-to-end learning for visual odometry
- GNSS positioning error prediction using neural networks
Test Platforms
- Ground vehicle equipped with GNSS, IMU, LiDAR, and cameras
- Fixed-wing and multirotor UAVs for aerial navigation testing
- ROS-based software stack for rapid algorithm development
Applications
Our research directly enables technologies for:
- Level 4/5 autonomous driving
- Last-mile delivery drones
- Urban air mobility (UAM) vehicles
- Precision agriculture UAVs