Smarter sensor processing for UAS payloads

2UAS payload designers are packing more and more processing capability behind the sensor, enabling more data filtering at the sensor level so only the most important information gets to the commanders in the field, cutting down on communication bottlenecks.

Unmanned Aircraft Systems (UASs) are still a bright spot for U.S. defense technology, even in light of the shrinking Department of Defense (DoD) budget. Despite the economic pressure, the U.S. still needs Intelligence, Surveillance, and Reconnaissance (ISR) capability operating around the globe night and day to keep an eye on threats and, where necessary, to remove those threats. UASs are their best tool for doing so today – from small, hand-thrown aircraft to large systems such as the Predator C Avenger and Global Hawk.

While these programs are facing budget pressures, it is milder than some other platforms due to their cost effectiveness and mission necessity. Integrators of these systems are taking advantage of Commercial Off-the-Shelf (COTS) computing technology to help alleviate some cost issues while leveraging the COTS performance capability to enhance UAS ISR payloads.

“The goal is to support a smarter sensor for sophisticated ISR payloads in tactical UASs,” says Tom Roberts, Solutions Marketing Manager at Mercury Computer Systems in Chelmsford, MA. “With new graphics processors such as those from NVIDIA and AMD, we are putting as much processing power as possible behind the sensor, enabling it to mine a sea of information, so that the 99 percent of data that is not needed is not transmitted to ground – just the needles in the haystack they want to find.”

“The primary goal of a UAS is to provide real-time imagery and intelligence, but if the radio link is weak they sometimes end up making a compromise with compression and even distribution,” says Robert Kubis, Product Marketing Manager at FLIR in Wilsonville, OR. In many cases, a direct line of sight downlink is necessary to get the highest-quality imagery, he continues. Beyond line of sight can be limited when it comes to image detail and fidelity. “They are trying to extract the intelligence before its loss due to the radio link.”

“It has become a recurring customer mantra: ‘We want more capability than we had previously – but using less Size, Weight, and Power (SWaP) than the older systems used to,’” says Peter Thompson, System Architect, Military and Aerospace, GE Intelligent Platforms. “For ISR systems, this is driven by increasing focal plane array size, faster frame rates, and more sensors of different modalities. We see demand for taking existing functionality – say a target tracker or a video compression unit – and dramatically reducing the SWaP of the hardware to fulfill that need.” GE offers an automatic video tracking solution called the ADEPT3000.

“A lot of UAS folks use cost-effective, lightweight sensors and camera cores for the smaller aircraft, whereas the bigger platforms can be more complicated,” FLIR’s Kubis says. “Even in manned aircraft, customers want to do more with their payloads – such as through imaging exploitation techniques to capture more data with cameras. FLIR makes use of many different high-end electronics components in their payloads to meet these performance demands such as FPGAs, DSPs, image processing techniques, and focal plane technology. We also push the envelope and talk about lowering SWaP and how can we do more in the product to manage thermal issues.” FLIR’s Star SAFIRE 380HD system flies on larger UAS platforms, and its BRITE STAR II currently is on the Fire Scout UAS from Northrop Grumman. FLIR cameras are also on numerous small UASs such as the Raven.

21
Sidebar 1
(Click graphic to zoom)

High-performance computing in the payload

“The big trend is to get as much processing power into the payload as possible, which creates tougher size, weight, and power challenges for the designer,” says Dinesh Jain, Senior Product Manager at Mercury Computer Systems. “They need to figure out how to extract the maximum performance at the lowest power. This can be done through delivering heterogeneous computing and dividing up tasks among the devices best suited for particular applications – such as DSPs, FPGAs, and GPGPUs.

“Heterogeneous computing is where you have very specific processors assigned specific tasks,” says Scott Thieret, Technical Director at Mercury Computer Systems. Mercury’s StreamDirect technology helps enable this process. The technology allows “users to transfer data from the sensor via Serial RapidIO and bypass the processor, transferring directly to the GPU. This improves overall efficiency and also cuts down on the design footprint, which cuts down on power consumption.”

“Heterogeneous computing also can help with multisensor payload designs,” Roberts says. “Military payload designers more and more are looking to fuse the data coming in from multiple sensors such as cameras, radar, etc., into one image that enables the ground analysts to get a better understanding of what is happening in theater.”

“When designing sensor payloads today, we need to think in terms of a network of sensors that [has] a hardware architecture and a software architecture,” says Eran Strod, Systems Architect at Curtiss-Wright Controls Defense Solutions. “Today ISR sensor architectures take advantage of the intensive DSP computing capability of Intel GPUs and other processors.” For more on Curtiss-Wright High Performance Embedded Computers, visit www.cwcdefense.com/products/subsystems/program-specific-subsystems/hpec-software-platform.html.

“However, the software architecture is just as important and must be able to have all the applications talk to each other,” Strod continues. “The software/hardware collaboration increases the flexibility of being able to talk to other entities or people on ground or other applications or the continental U.S. All of that stuff requires middleware, and Intel is the best platform from an enterprise software perspective. Intel has the most ubiquitous software story as it unlocks openness and functionality. The benefit of having more openness than ever on the software architecture side combined with high-performance number-crunching engines is really exciting from a design standpoint.”

Open architectures also enable rapid deployment of new payloads. Along those lines, engineers at General Atomics Aeronautical Systems, Inc. (GA-ASI) in San Diego and SELEX Galileo have developed an open payload architecture concept for Predator B unmanned aircraft that enables payload suppliers and mission systems integrators to design their own payload control software, then integrate their own payloads, according to a GA-ASI release. GA-ASI engineers handled the hardware and software modifications to the Predator B system, and SELEX Galileo provided the radar during a recent demonstration of the concept.

“We continue to see new opportunities for COTS products, primarily because the defense contractors and system integrators don’t have the resources to develop their own hardware and/or they want to focus their engineering resources on their value-add rather than reinventing the wheel,” says Rob Scidmore, President and CEO, Extreme Engineering Solutions (X-ES) in Middleton, WI. The COTS XPand6000 Series from X-ES uses existing rugged COTS COM Express modules for the processing subsystem and rugged COTS PMCs or XMCs for the application I/O, he adds.

Small UASs and small form factors

“The smaller payloads do obviously have lower size, weight, and power requirements,” says Michele Evans, Vice President, Lockheed Martin MS2 Business Development. “Power efficiency is important with the smaller aircraft, as you have to weigh extra battery power versus the speed and altitude requirements. If a particular mission requires greater speed and altitude, the payload will have to be less weight and lower power.”

“Payloads for small, hand-thrown UASs such as the Lockheed Martin Desert Hawk (Figure 1) – produced for the United Kingdom – can basically be interchanged by the warfighters themselves, Evans says. “The payloads are fairly small, and installing them is a pretty simple procedure where the warfighter just unscrews four bolts, then swaps in the next payload. The small UAS uses a typical Electro-Optical Infrared (EOIR) payload that uses basic cameras, so ground troops can see what is ahead of them in night or day. Future sensor payloads may also include a miniature synthetic aperture radar.

21
Figure 1: The Desert Hawk hand-thrown UAS has a payload that warfighters can swap out just by removing a few bolts.
(Click graphic to zoom by 1.9x)

Keeping it cool on a UAS

“The lower SWaP requirements present design challenges in UAS environments because of the cooling constraints in unpressurized high-altitude environments,” says Andy Mason, MilAero Products Business Manager at Kontron in San Diego. “Designs also are trending more toward reduced SWaP COTS system solutions. We have some small form factor products in use on UASs. One is the small form factor embedded Cobalt computer, which takes advantage of high-speed processors in a small conduction-cooled box.”

“For some unmanned applications, the size and cooling requirements are pretty tight, so solutions such as 3U VPX in a conduction-cooled half ATR box can be used with or without fans, depending on the mission requirements,” Mason says. “There are some challenges in cooling small form factor designs. One is the desire designers have to avoid fans for increased reliability and reductions in cost and size. Some are going with a natural convection approach that limits the amount of thermal energy you can dissipate. Effective systems making use of the heat sink can cool up to 50 to 100 watts, but beyond that will need help such as adding a cold plate for additional cooling.”