Operations in Degraded Visual Environments using Millimeter Wave Imagery

Corvid Technologies and the University of North Carolina at Charlotte propose to develop Clear Path Hazard Avoidance Radar (CPHAR). CPHAR is designed to enhance situational awareness in degraded visual environments (DVE) by utilizing millimeter wave (mmW) radar imaging and advanced computer vision techniques. CPHAR will address the limitations of existing systems, such as visual cameras, night vision cameras, and LiDar, which struggle to see through low light, fog, dust, rain, snow, fire, smoke and other visual obscurants.The team designs a practical, novel, and economic mmW sensor package composed of commercial off the shelf (COTS) components. We collect CPHAR model training data using a data collection rig that simultaneously records mmW imaging data and other forms of data for automated data labeling using cross-modal supervision techniques. Reliable and robust models are trained by collecting a large dataset and augmenting the dataset with additional real-world data collection and synthetic data generation in an iterative informed sampling training loop. Finally, an intuitive human-machine interface displays an ego-view based on mmW imaging data that mitigates the effects of DVE.