Intel has partnered with Daedalean, a Swiss startup developing machine learned solutions for the aviation industry. Their latest white paper presents a reference design for an AI application that acts as a never-distracted co-pilot and is certifiable, meaning it meets regulatory testing. By publishing this whitepaper, Daedalean and Intel hope to provide guidance to other companies looking to integrate certifiable machine-learned electronics and applications into their aircraft.
Debra Aubrey is Technical Product Marketing Manager at Intel Corporation.
“The aerospace industry still needs the first step towards a future with multi-directional embedded computing equipment: a reference architecture or a specific list of requirements to develop the right types of computers,” she said. “A reference architecture includes regulatory requirements, low-level and high-level software, and silicon solutions for machine-learned applications. Regulatory agencies must review a reference architecture to certify that it provides predictable, safe behavior in the sky.”
Daedalean has been working on a machine learning algorithm and reference architecture for a computer that can run it. They tested the reference architecture in labs and on board aircraft to develop situational intelligence, the ability for machine-learned applications to predict and respond to future events. To reduce time-to-market for companies interested in their applications, Daedalean has partnered with Intel to supply silicon to manufacture these applications. The two companies worked together on a reference architecture that will reduce time to market and enable companies to more quickly integrate machine-learned computers into their cockpits.
The white paper lays out the reference architecture for certifiable embedded electronics, including the challenges of applying Software Assurance to machine-learned devices, the visual awareness system they use, and the current and future role of embedded computing in the industry. The report also looks at the software and hardware requirements that ensure aviation systems are safe and effective.
According to a statement from Intel and Daedalean, the reference architecture can “improve time to market for organizations interested in integrating what has shaped them, situational intelligence — the ability to not only understand and comprehend the current environment and situation, but also anticipate a future situation and react to it – in the cockpit.”
dr Niels Haandbaek is Technical Director at Daedalean.
“This is the first document ever that presents a real working example and provides guidance on how to address the challenges of implementing machine learning application in flightable embedded systems in general: how to ensure that your ML-based system meets the meets computational requirements, certification requirements and the size, weight and power (SWaP) limitations simultaneously. The approach outlined in the document is driving the aviation industry’s need for high-performance embedded computing,” he said.
This white paper can help bring the power of AI to avionics. It is the first document to present a working example of a machine-learned system and provide guidance on overcoming application challenges. The actionable recommendations and insights in the new report can fuel the industry’s desire for high-performance embedded computing. This fundamental real-world example has the potential to create a new wave of flyable machine-learned applications.
You can download the white paper here.
