As time passes, inventions tend to become ingrained in our everyday lives and less magical once the novelty wears off. I imagine people from the 1920s considered inventions and innovations of that time pretty amazing. Widespread availability of electricity brought to life radio broadcasts, traffic signals and television. As these wonders of technology have always been a part of my life, admittedly, I take them for granted and don’t have the same level of appreciation for them as someone seeing or hearing them for the first time.
Fast forward to 2020. I very well may become one of those who marvels at what will be one of mankind’s greatest achievements—a truly autonomous flying vehicle. Like many innovations of the past, there are a lot of moving parts that must work seamlessly to successfully–and safely– launch and sustain a flying vehicle.
From a technology standpoint, the processing capability to capture, store, manipulate, and disseminate the sensor data necessary to launch, fly and land an autonomous vehicle is available today. It’s when introducing other elements—passengers, stationary object avoidance, other flying cars —that safety-critical performance becomes a focus. The onboard processing demands of smarter, more integrated avionics applications such as urban air mobility (UAM) platforms (e.g., a flying car) are overloading the single-core processors traditionally used to meet safety-certifiable computing requirements.
Multicore processors, first introduced in 2001, have found footing in virtually every aspect of computing applications, yet their adoption has been slow in those applications requiring safety certification. Designed by silicon vendors to improve average core performance, multicore processors are designed to share resources, introducing non-determinism (algorithms that, even for the same input, can exhibit different behaviors on different runs) and control interference. Flight safety is reliant on computing systems performing the same action every time within a given timespan when the system encounters a specific scenario. Non-determinism and control interference inject disruption and complexity to the safety decision process, making DO-254 (RTCA) or DAL (Design Assurance Levels) certification challenging.
Mercury solved this challenge by working closely with Intel to design a powerful, multicore processing solution with deliverable design assurance level artifacts for mission-critical, flight safety-certifiable applications. Intel and Mercury’s design and flight safety certification experts addressed the increasing performance demands of smarter and more integrated avionics applications by developing the award-winning CIOE-1390 COM Express module. Powered by Intel Atom® multicore processors, the CIOE-1390 provides a full x86 processing architecture and on-die GPUs featuring Intel’s latest graphics. Supported with BuiltSAFE™ technology, the CIOE-1390 modules are available initially with DO-254 DAL-C flight safety certification evidence for the circuit card assembly and DO-178C DAL-C evidence for the highly optimized custom BIOS and bootloader software. The availability of these artifacts enables system safety certification to be performed faster, at a lower cost and with less risk than by other approaches. Only the size of a credit card, the rugged CIOE-1390 module delivers the processing power required to solve complex mission-critical compute problems that help ensure an autonomous flying vehicle and other graphic-intensive avionics applications have the onboard processing muscle needed to perform, flawlessly.
I’m intrigued at the prospect of witnessing a sky highway populated with driverless vehicles. Even more so, knowing that through Mercury’s commitment to Innovation That Matters® we are delivering the technology solutions critical to making the world a safer, more secure world for all. Beep, Beep.