The Quest for Faster, Lighter and Cheaper Next-Gen UAVs: Q&A with General Micro Systems
Recognizing the role Intel architecture is playing to assure UAVs—including small UAVs—are no stranger to SWaP-C.
Not long after ABI Research reported: “Increasingly favorable regulatory environments expand the roles for small drones, also known as small unmanned aerial vehicles (sUAVs), in commercial applications,” EECatalog asked GMS CTO Chris A. Ciufo what this information means for the computer manufacturer. Edited excerpts of Ciufo’s response to that question and others follow:
Chris A. Ciufo, General Micro Systems: When considering size, weight, and the amount of processing it’s possible to pack into a “small” platform you also have to consider what’s meant by “small.” When VME became the dominant form factor in deployed defense systems, it was in the roughly 6 x 9 inch 6U form factor—which is anything but “small” in a world of smartphones and IoT devices. But VME’s successor—VPX in 3U size—is “sort of small” at roughly half the size of VME. Rugged boxes based upon 3U VPX might possibly be deployed in UAVs and sUAVs if they can tolerate a half-a-shoebox weighing 30-50 pounds. I’m skeptical.
Also, despite the promise of “openness” with 3U VPX, it’s really not—as vendors’ cards are not interchangeable. So why not choose a system that is completely designed for the task at hand and SWaP-C optimized for UAVs? Already in use by our customers in air vehicles, our most svelte small form factor modules are credit card-sized, based upon industry-standard Intel® PC architectures and its newest processors, and available in rugged, conduction-cooled chassis not much bigger than a pack of Las Vegas playing cards and weighing 1.5 pounds (Figure 1).
EECatalog: How will the UAV market evolve over the next five years and how well has your company’s experience and focus prepared it for this evolving market?
Ciufo, General Micro Systems: The focus for UAVs is going to be faster, lighter, cheaper. What one can do in a smartphone today would’ve required several PC-like systems 10 years ago—plus the RF sections. Even today’s small, rugged, efficient boxes like our “SB” series of Intel® Core™ i7/Xeon® products may prove to be too big when compared to what’s in a smartphone.
Yet smartphones are designed for 10 million unit volumes, and DoD UAV systems are at what, 100-piece volumes? So cost- and size-optimizing will be impossible compared to a smartphone—and don’t forget the requirement for exceptionally robust MIL-STD circular connectors. Those things are huge and constrain our chassis sizes. UAV and sUAV systems will get smaller, but there is an economical limit to how small they can get for this particular (defense) market.
Still, the IoT is making small, powerful components available in smaller volumes that might be perfect for next-gen smaller UAVs. We’re closely looking at some of Intel’s IoT processors and SoCs, which have the benefit of the widely used x86 architecture (and all of the software available for it) but are in dramatically reduced IoT-sized form factors. New form factor rugged boxes—perhaps even smaller than our Green series products, which are credit-card-sized—are one way to insert IoT technologies into UAVs. As the physical size gets smaller, so must the cables and interconnects. It’s likely the high-pin-count military-style connectors will have to go, replaced by as-yet-unknown commercial-style cables.
At the same time, some systems may be unable to accept these kinds of COTS technologies. In that event, we will continue to evolve and shrink our modular small form factor systems, changing out our “engines,” adding new features, but still offering the tried-and-true military-style packaging and connectors.
EECatalog: What role do you see QoS guarantees playing as advances in video recording and drone technology continue?
Ciufo, General Micro Systems: Higher resolution, isochronous data, and guaranteed “pipeline” bandwidth are all important when video imagery is being used for decision-making, whether in real time or for offline analysis. So guaranteeing the quality of the data or available bandwidth are nice-to-have features, especially as battlefield networks get loaded up and congested. Yet UAV sensor data sent to ground stations or operator consoles are typically 1:1 (dedicated), so only interference or other RF obstacles are the anomalies reducing that quality.
QoS is a nice-to-have when sensor data and imagery is being transported over wider networks but not as essential when dedicated pipes are available. In our case, we have systems that are data-optimized to assure maximum throughput with minimum latency. One imaging system used by the U.S. Army requires that we provide 30 frames/sec of real-time data with a latency target of around 0.03 second: this is non-trivial and requires extensive DSP, but we maintain QoS via high-performance hardware (Figure 2).
EECatalog: How can imaging systems such as the one you describe work with Augumented Reality and what tradeoffs need to be considered?
Ciufo, General Micro Systems: It would be helpful to mount the imaging system on a ground vehicle and to make it possible for the system to take multi-spectral sensor data in real time. Our imaging system uses very specialized Army software and algorithms to “fuse” the images and data together and displays it in (near) real-time with almost zero latency. This is so operators can control their vehicles in an Augmented Reality environment even if the view out the windows is completely obscured by darkness, smoke or a layer of dirt caked across the glass. Clearly such a system can be created—we’ve done it with the Army—but at every step of the way the challenge was trading off system complexity and price against the performance required for real time. In the end, we did this with two boxes: a high-performance server and video processor, plus a video encoder and sensors interface. The system is small, cost effective, and high performance in a small, mobile space. No fans are required: it’s all conduction-cooled.
EECatalog: How can the progress of technologies that allow UAVs to fly in civilian airspace, e.g., collision avoidance algorithms, open additional markets for the application of unmanned aerial vehicle technology?
Ciufo, General Micro Systems: I’m afraid this is outside of my area of expertise. But these kinds of COTS technologies—developed for the consumer drone market and benefiting companies like Amazon that want to deliver products via drones—will have direct benefit on defense-related UAVs, as well. Unless during wartime or via special exception, defense UAVs that fly via operator control in civilian airspace need the same clearance from Air Traffic Control as commercial airliners. This is for the safety of civilians in the air and on the ground, and for the success and protection of the UAV itself. If this kind of COTS technology were readily available and added to DoD platforms, I would imagine the separation rules and clearance requirements might change. The FAA already created and implemented the Small UAS Rule in August 2016, so the agency is being remarkably nimble as technology evolves.
A minor footnote: if collision avoidance can be added to DoD UAV airframes—then smaller sensor platforms that are completely autonomous might be developed for reconnaissance, interdiction, or ISR/EW purposes. That is: we can create smaller and more capable sensor platforms that are more sneaky to get the job done.
EECatalog: How is GMS positioned to support a couple of technologies likely to move forward together: video capture and unmanned flight?
Ciufo, General Micro Systems: General Micro Systems has a 37-year history of creating exceptionally rugged, lightweight, and creatively feature-packed defense systems that evolve with the latest technology. In fact, our core intellectual property (IP) is our modular architecture that is intended to be upgraded. We helped create and make VME a success, while today we specialize in rugged small form factor systems. As the industry went from Motorola- to Intel-based processors, we evolved as well. We’ve always taken the latest technology and quickly deployed it to the warfighter. We’ll keep evolving.
With these UAV technologies including video capture and autonomous/unmanned flight (control systems) we are already finding ways to evolve our mission processor platforms and airborne network servers. VR/AR/machine vision technologies like Intel’s RealSense™ cameras, which were featured heavily at this year’s Consumer Electronics Show, have utility not only on the airframe/vehicle, but in the operator’s console and control station. Think about the ability to use in-air gestures to control a vehicle in flight, or if autonomous, interact with the vehicle’s data stream without being constrained by having to push buttons. Faster decisions can be made with bigger sensor data sets: this saves lives and improves mission success.
Regardless of how the technology evolves, I’m confident that General Micro Systems will find use for it and bring it to the battlefield quickly to improve mission success.