jnz/INSLIB
Open Source Inertial Navigation Library
What it solves
INSLIB provides a reliable way to calculate the 3D position, velocity, and orientation of moving objects—such as drones, cars, or robots—by combining data from multiple sensors. It solves the problem of "noisy" or incomplete sensor data by using Kalman filters to fuse inputs like GPS, accelerometers, gyroscopes, and barometers into a single, accurate navigation state.
How it works
The library uses a set of Kalman filters to process sensor measurements. It is written in portable C11 and is designed to be "embedded-friendly," meaning it runs efficiently on microcontrollers without requiring an operating system or heap memory. It compensates for common issues like sensor latency (the delay between a GPS reading and when it is processed) and handles sensor outages by relying on inertial data when external references like GPS are temporarily unavailable.
Who it’s for
It is built for engineers and developers working on robotics, aerospace, or automotive systems who need precise, real-time navigation tracking. It is particularly useful for those working on resource-constrained embedded hardware who need a battle-tested, high-performance solution that can be easily integrated into existing projects.
Highlights
- High Reliability: Features over 90% test coverage, including MC/DC testing, and follows aerospace-style requirements traceability.
- Embedded Optimized: No heap usage, no OS dependencies, and known worst-case stack usage for every function.
- Robust Filtering: Uses UDU/Bierman-Thornton Kalman filters for numerical stability.
- Flexible Integration: Includes a variety of modes, from simple roll/pitch estimation to full 3D navigation, and provides Python tools for post-processing and data visualization.
- Hardware Agnostic: Works on both bare-metal embedded targets and desktop computers.
Related
- Project
- Project
- Project
- Project
- Project