AI VisionCube DT

The AI VisionCube DT combines two visible-light cameras (short-focus 3.9mm + long-focus 12mm) and a 384×288 thermal imaging module on a single 6 TOPS image-processing board. The result is the most versatile mid-tier configuration in the AERVUE AI Vision range: 1.2km long-range detection across the visible spectrum plus 384 thermal sensing for round-the-clock UAV operations under any lighting condition.

All three sensors feed into the same YOLOv7-derived multi-target processing architecture, identifying up to 50 dynamic targets per frame and locking onto human and vehicle subjects through crosshair auto-lock, Kalman trajectory memory, and adaptive snap-tracking. The high-speed lock success rate reaches 92%, and re-acquisition of briefly occluded targets completes within three seconds — across visible OR thermal channels.

Picture-in-picture architecture displays full-scene wide-field imagery in the main view while overlaying 8× zoomed detail or thermal in the sub-view, enabling pilots to maintain situational awareness while tracking specific targets at maximum sensor performance. Camera switching between short-focus visible, long-focus visible, and 384 thermal happens via a single remote-control channel toggle.

The AI VisionCube DT runs on 9–16V input, communicates via CRSF, supports BetaFlight or ArduPilot, and integrates into 7-inch or larger UAV airframes in under five minutes. End-to-end video latency stays below 60ms across all three sensors, total kit weight remains under 200 grams, and operating temperature spans −20°C to 60°C — extending the platform’s mission profile into search-and-rescue, perimeter security, and infrastructure inspection workflows that demand multi-spectral coverage.