Vision Based Autonomy for Group 1-3 Aircraft in GPS-denied Environments

KEF will mature our 'aerial autonomy subsystem' to meet Air Force customer and end user needs for aerial autonomy. The primary technical objectives are to adapt KEF’s existing autonomy solution to: (1) Showcase the ability to obtain a navigational position fix through visual terrain feature recognition in GPS denied environments (also known as ‘Terrain Relative Navigation’ and (2) Upgrade and deploy the autonomy software on Air Force relevant sensing and computing (e.g. monocular cameras, US-built processors). KEF autonomy--which includes GPS-denied navigation, hazard detection and avoidance, path planning, and machine-learning target localization--has reached TRL 6 for Group 1 sUAS within indoor and outdoor environments. In Ph. 2, we seek to extend these capabilities to new flight regimes, specifically to higher altitudes, up to 400 feet on multi-rotors and up to 10,000 feet on fixed wing platforms. We will also develop new models for gimballed, rolling shutter cameras and improve the scalability of our autonomy system, allowing it to be rapidly deployed to new Air Force relevant aircraft. These adaptations are assessed in simulation and field tests on small multirotor UAS as well as high altitude open-loop tests on piloted aircraft. For the past nine months, KEF has worked with a DOD customer to deploy our autonomy subsystem on a class 1 government-off-the-shelf (GOTS) multicopter sUAS, “Deep Purple”.  These DOD customers have provided us with insights into technical considerations, including air worthiness permitting, supply chain considerations (e.g. non-Chinese electronics), and operational environment considerations.