Overview
An embedded navigation computer combining GNSS positioning with inertial sensing to produce a continuous, drift-corrected state estimate suitable for small autonomous vehicles.
Architecture
The board integrates a GNSS receiver, a 6-axis IMU, and a dedicated microcontroller running a real-time state estimation loop. Output is streamed over a serial link in a compact binary format.
Design
The core design goal is graceful degradation: the system should produce a usable estimate even during GNSS outages, relying on inertial dead-reckoning until a fix is reacquired.
Implementation
State estimation is implemented as an error-state Kalman filter running at a fixed update rate, with GNSS measurements incorporated asynchronously as they arrive.
Experiments
Bench testing compares filter output against a reference GNSS log to characterize drift during simulated outages of varying length.
Results
Short outages (under 30 seconds) show minimal positional drift; longer outages expose the need for additional aiding sensors, motivating future DVL or odometry integration.