Flare Bright has developed a patent pending, Machine Learning (ML) enhanced software system that boosts the performance of inexpensive Inertial Measurement Units (IMU), resulting in a degree of accuracy that creates an effective low cost, low weight and low volume Inertial Navigation System (INS) for extended flight operations of unmanned aerial systems (UAS) in GNSS-denied environments.

This enhancement can give highly accurate position data when GNSS fails, adding redundancy to flight safety critical systems, and is widely applicable within integrated navigation performance management systems on a wide range of drones and aircraft.

Flare Bright has previously presented preliminary flight tests results of this GNSS-free capability gathered using a 1.2m wingspan in-house fixed wing drone at ENC2023. In this paper, Flare Bright will present new data gathered by deploying our solution in a realistic operational scenario using a representative 2m wingspan fixed wing operational drone over multiple terrains, including over water where visual navigation is not possible (Fig 1).

The results, from flights up-to 1 hour long, will show how Flare Bright’s model-based navigation software enables a $5 smartphone IMU to outperform a $10k tactical grade IMU (HG1930) in approximately 20 minutes, and results in an average performance approximately 4 times the error of the $100k navigation gold standard IMU HG9900 after 1 hour (Fig 2).

Flight test data will be analysed in conjunction with simulation data to provide a critical review of the constraints and potential capability achievable with this novel technique. The results presented will demonstrate a credible route towards a practical, operational, future capability within the emerging UAS sector and the potential value of using mass-market sensors with software enhancements within the wider aviation sector where safety is paramount.