104 on a proprietary waveform that enables scalable, self-forming and self-healing mobile ad hoc networks. It is Mesh Rider’s operation in multiple frequency bands (up to six in one radio) combined with spectrum-sensing capability to detect RF interference and jamming, and to switch to another channel or band. While Black Widow is offered with a choice of C-code or more robust and secure M-code GPS receivers, recent experience has shown that Russia, for example, can effectively deny GPS service in targeted areas, so the UAV uses visual-based navigation as a back-up. The vehicle’s two downward-facing cameras image the terrain beneath, and compare it with downloaded maps and elevation information, enabling it to determine its position in the absence of a usable GPS signal. Visual navigation The system in question is from partner Palantir and known as VNAV, and it runs alongside the GPS and inertial sensors. To work out the UAV’s position and movement, the algorithm takes in three main information components. The first is the feeds from onboard sensors such as the compass, accelerometers and barometric altimeter. The second is optical flow (visual inertial odometry), based on the feed from the downward-facing cameras, which the system watches frame by frame, applying mathematical transforms to the data to work out the vehicle’s velocity and position over time. For navigation over a short time, perhaps filling in for a localised GPS outage, that might be sufficient, but over longer periods small errors build into larger ones until the vehicle eventually becomes lost. Correcting for this drift is the job of the third component, which is Palantir’s reference-matching technology. This takes images from the camera system and matches them with pre-rendered satellite imagery in multiple wavelength bands, using computer vision to determine where those images align and finding the vehicle’s position from that. All data sources go through a Kalman filter to calculate the most accurate solution. Palantir keeps the imagery of the operational area up to date by tasking partner satellites and delivering it to the UAV systems over a variety of communication bearers. VNAV’s visual navigation algorithms are deployed directly onboard the uncrewed vehicle and run fully offline using edge runtime computing, also from Palantir. Running repairs Damage and mechanical failure are always a risk in hostile environments, so Red Cat has protocols that increase the chance of the vehicle being recovered, as Thompson explains: “In the event of a mechanical failure that does not fatally impede flight, the operator can command an autonomous return to home (RTH), February/March 2025 | Uncrewed Systems Technology Render of underside, showing battery pack, and both forward- and downward-facing cameras (the former for obstacle avoidance and the latter for visual navigation), and matching camera feeds with stored terrain imagery (Image courtesy of Red Cat) Palantir keeps the imagery of the area up to date by tasking partner satellites and delivering it to UAV systems over communication bearers
RkJQdWJsaXNoZXIy MjI2Mzk4