Obstacle Detection and Intelligent Notification (ODIN)

Soldiers in the field increasingly use mechanical pan-tilt-zoom (PTZ) video cameras to operate systems, whether from below the closed hatch of an infantry carrier vehicle or through the operator control unit of an Unmanned Ground Vehicle (UGV). However, mechanical PTZ cameras force the user to choose between wide-angle imagery that provides a valuable big picture, but contains insufficient detail to detect or recognize distant objects, or narrow-angle imagery that captures the details needed to identify important objects (e.g., obstacles), but removes key contextual information, reducing situational awareness and making visual search difficult and fatiguing. Our proposed solution is a mobility aid for performing Obstacle Detection and Intelligent Notification (ODIN). The system detects and tracks potential obstacles in high-resolution (multi-mega-pixel) monocular video, produces region-of-interest cues and threat rankings, and presents the information to one or more users via an intuitive Virtual Pan-Tilt-Zoom (VPTZ) framework. ODIN leverages our existing obstacle detection and tracking technologies together with novel feature extraction and search strategies to achieve real-time detection and tracking of potential obstacles at long distances, and is largely hardware-agnostic.