Low-Altitude Sensor Explotation

Intelligence products generated from sensors currently deployed on U.S. Army Airborne ISR platforms only utilize data from a single sensor modality. As such these products are limited in both current capability and the ability to extend to new intelligence missions. Toyon Research Corporation has developed algorithms to fuse multiple image bands and LIDAR and perform AiTR using deep neural networks. On this effort Toyon proposes to transition fusion and deep learning algorithms from high-altitude sensors to low-altitude sensors in the Full Spectrum Targeting system for the Future Vertical Lift (FVL). Processing data from the Full Spectrum Targeting system requires a number of adaptations to our previously developed algorithms including consideration of new spectral bands (SWIR, MWIR, LWIR, NIR), development of new transfer learning and training methods to create band specific feature representations, adaptation our LIDAR processing for the FVL integrated LIDAR system and adaption of the AiTR architecture from high-altitude sensor perspectives to low-altitude perspectives.