Navigation System
The autonomous navigation system is the core of Pangolin AT-1's intelligence. Fully developed in-house from hardware to algorithms, it delivers reliable centimeter-level positioning across both indoor and outdoor environments.
Outdoor Navigation: RTK + Global Planning
RTK Positioning
Real-Time Kinematic (RTK) differential correction provides centimeter-level absolute positioning:
- Base Station: Receives satellite signals at a known fixed location
- Rover (Vehicle): Receives both satellite signals and correction data from the base station
- Accuracy: Horizontal ±0.8 cm, vertical ±1.5 cm (open sky conditions)
- Update Rate: Up to 20 Hz for real-time position tracking
Path Planning
The global path planner automatically computes optimal routes:
- A* and Dijkstra-based algorithms for shortest-path computation
- Terrain cost maps integrate traversability analysis
- Multi-point cruise and timed loop operation modes
- Dynamic re-planning when obstacles block the original path
Indoor Navigation: Laser SLAM
SLAM Mapping
Simultaneous Localization and Mapping (SLAM) enables operation in GPS-denied environments:
- No Pre-Installed Infrastructure: No beacons, markers, or reflectors required
- 2D/3D Mapping: Choose between efficient 2D grid maps or detailed 3D point clouds
- Map Editing: Modify, annotate, and manage multiple maps in the control terminal
- Multi-Map Support: Store and switch between maps for different floors, buildings, or zones
Indoor Positioning Accuracy
| Environment | Typical Accuracy |
|---|---|
| Structured indoor (warehouse, factory) | ±0.5 cm |
| Semi-structured (parking garage, hallway) | ±1.0 cm |
| Dynamic environments (with moving objects) | ±2.5 cm |
Multi-Sensor Obstacle Avoidance
Sensor Fusion Stack
- 3D LiDAR: Primary obstacle detection, 360° horizontal FOV, range up to 30 m
- Depth Camera: Close-range supplement, excels at detecting low obstacles, glass, and overhanging objects
- IMU: Inertial measurement for heading, tilt, and acceleration compensation
- Wheel Odometry: Dead reckoning fused with visual and inertial data
Dynamic Obstacle Handling
- Real-time detection and classification of moving objects
- Predictive trajectory estimation for crossing objects
- Safe stopping distance maintained based on vehicle speed
- Alternative path computation when path is blocked
Indoor-Outdoor Seamless Switching
The positioning fusion engine monitors signal quality from RTK and SLAM sources simultaneously. When transitioning between environments (e.g., driving from a warehouse through a door to an outdoor yard), the system:
- Detects degrading RTK fix quality as the vehicle moves indoors
- Smoothly increases SLAM weighting while decreasing RTK weighting
- Maintains continuous position estimate during transition
- Reverses the process when moving back outdoors
This transition is fully automatic and requires no operator intervention.