At Daedalean, we aim to show the world that artificial intelligence, far from just a trending buzzword, describes a set of technologies that will in the foreseeable future raise levels of safety and efficiency in aviation to new heights. AI is a feasible, inevitable advancement. And Daedalean is focused and determined to reveal to the world that among these technologies are some that are already well-harnessed, mature, and ready to move out of the confines of the lab and finally to take flight.
“Looking ahead to my first full year leading Daedalean, I’m excited to report we will see the fruition of several years’ labor in the coming months and years,” stated Bas Gouverneur. “This will be a year full of challenges, certainly, but also a year of accomplishments. I joined the company at a decisive moment, and I am eager to see all the promise of the indispensable utility of AI in safety-critical applications to be delivered in the very near future, with the real-world examples demonstrated, and doors to its certification opened for all. We are at a pivotal moment in the history of aviation, and, indeed, in the history of technology.”
Neither panacea nor dystopia—just better
Some forecast the adoption of AI will bring society’s downfall. Innovation has, for centuries, continued to lift us up.
From the printing press to electricity to the automobile to the airplane, dramatic technological innovation has a long history of resistance. The Luddites’ violent destruction of power looms stands as perhaps the most enduring example of misguided opposition to life-improving change. Indeed, few would advocate a return to the tedium, time sink, and poor quality the Luddites sought to maintain. Certainly none would suggest current standards of living could be sustained by guarding that status quo. Better machines made weaving better.
The sudden entry of AI into our everyday conversations can feel shocking. But machine learning, the type of AI Daedalean develops, is, basically, numbers. Machine learning uses statistics to make sense of mountains of data very quickly. Machine learning makes machines better. And those machines make flying safer.
According to an IATA safety report, lack of situational awareness is a contributing factor in 33% of accidents—that’s where Daedalean can make a difference. We can improve a pilot’s situational awareness, and provide something even better: Situational Intelligence. Combining sensors with onboard processing power, we can offer the ability to not only understand and make sense of the current environment and situation but also anticipate and react to a future threat. Achieving full situational intelligence is a ways off, but even today AI delivers huge improvements in awareness which translates into significant improvements in safety.
AI down low
Daedalean’s first products assist pilot perception to identify aerial and ground-based objects, navigate without GPS, and safely land. Together, these capabilities combine to serve one of the most dangerous and vital areas of aviation: low-altitude, vertical flight operations.
At high altitudes, such as those maintained by large commercial airliners, flight is very highly automated. The skies far above are relatively uncongested and most aerial traffic broadcasts identifying and positioning information via transponders. But at lower altitudes, both ground-based and aerial objects create a high risk tapestry of danger from wires to cranes to terrain to other low-flying aircraft. A set of aircraft-mounted cameras, scanning and analyzing the airspace for threats creates another layer of safety atop that provided by skilled pilots. And we have never before in the history of flight faced a threat we are now on the cusp of having to confront. It is a new, and fast-growing danger to low altitude flight: drones.
Drones on their own
More than 2,000 drone sightings have been reported near airports since 2021. Sixty drone incidents have forced pilots (including commercial aircraft) to take evasive action. Earlier this year, a drone collided with a Canadair CL-415 aerial firefighter assisting with wildfires in Los Angeles causing significant wing damage. One way to avoid collisions is to equip aircraft with cameras and a computer to detect drones. But drones should also take responsibility for their own safety.
Flying a drone safely beyond the visual line of sight (BVLOS) is one of the major hurdles to introducing unmanned aerial vehicles into the national airspace. And BVLOS operations would unlock the full potential of drones for crucial applications such as search and rescue, emergency cargo delivery, and natural disaster tracking. But at present, the national airspace is not equipped to handle the millions of drones projected to soon take to the skies. To combat the threat drones pose to low altitude flight safety, Daedalean is partnering with Swiss drone manufacturer Xer Technologies AG to demonstrate how vision-based systems can provide aerial safety for all aerial vehicles.
Xer and Daedalean have equipped an Xer X8 drone with Daedalean’s Ailumina Vista visual traffic detection system. With this system onboard the drone, externally-mounted cameras will be able to detect and identify nearby traffic. Being able to see other aerial vehicles that could pose a threat is valuable on its own, and we have demonstrated the capability for several years. But later this year, we will demonstrate the ability to automatically execute avoidance maneuvers.
“Being able to detect and avoid aerial traffic will be a game-changer for drones that operate beyond visual line of sight,” stated Gouverneur. “Networks of unmanned aerial vehicles will be able to scale up. We will be able to reach those in need by piloted or unpiloted means. Whatever is faster. We will have options, trusting that all aircraft are capable of maintaining safe separation, avoiding obstacles, and communicating seamlessly with one another in real-time. This will enable efficient, reliable, and adaptive responses to emergencies, deliveries, or other critical operations, even in complex or congested airspace.”
But knowing where to fly and where not to fly is much more useful if you know where you are. Over-reliance on satellite-based navigation systems has made this sometimes difficult.
When GPS is dead or has misled
The last few years have shown a significant increase in instances of jamming and spoofing of GPS, and more broadly, GNSS. Daedalean has been developing and demonstrating a vision-based system for years, but those demonstrations have provided pilot assistance only—information displayed on a tablet interface informed a pilot as to his or her position. The efficacy of this pilot assistance has been proven out in our own testing and with partners such as Leonardo Helicopters and others. But this year, we will take a significant step forward by integrating the capability into an autopilot system.
As we announced (link to press release) last year, we are developing a roadmap with Moog to integrate Daedalean technology into the Genesys Avionics Suite™. A significant milestone on that roadmap will be met this year when Moog’s Bell OH-58A Kiowa will be equipped with the Ailumina Vista visual awareness system and integrated into a Genesys autopilot to perform a never-before-attempted demonstration.
“We are integrating the GNSS-denied navigation system in closed-loop control with the flight control system of the Bell OH-58A, allowing autopilot flight without GNSS,” stated Gouverneur. “And we’re showing how this capability can be immediately relevant in search and rescue missions.”
The helicopter will execute a low-altitude search pattern over unknown terrain without using GNSS, relying instead only on the features of the landscape below for navigation. The system will also automatically identify suitable off-site landing spots that meet mission requirements. Once a suitable landing site is found, the system will calculate its position and automatically upload those coordinates as a new waypoint in the mission plan in order to execute a fully automated landing.
Such functionalities are reducing pilot's workload significantly and could prove indispensable in humanitarian and rescue operations when ground- and satellite-based infrastructure is compromised.
“Navigation in environments where GNSS is under attack has become a high priority across nearly all sectors of the aviation industry,” stated Gouverneur. “We have prioritized bringing this capability to users in line with our commitment to increasing safety in the skies.”
But what can be implemented onboard as soon as it's technologically feasible should not be hastily put into operation. A deeper sense of trust must be created first. So in aviation, where there is zero tolerance for failure, we have a rigorous and painstaking process for all new aspects of flight. It’s called certification.
The STC is the key
The regulatory green light takes many forms. Airframes, engines, propellers, and avionics are certified for airworthiness. The design of computer software and hardware is approved to meet rigorous levels of reliability, functionality, and cybersecurity. Regulatory oversight ensures that the airframe and every component and system, whether mechanical, electrical, or digital, operates within established guidelines to protect lives and property. In aviation, certification and approval are not just technical requirements but the cornerstone of trust, reassuring passengers, operators, and regulators that every aspect of an aircraft and flight operations meets the highest safety and performance benchmarks. It is an ongoing process, evolving alongside advancements in technology and the changing demands of the industry.
We’ve outlined to global regulators how we believe systems benefiting from machine learning can be fit for purpose. Now, we are working our way through the regulators’ meticulous, multi-part process for a Supplemental Type Certificate (STC) for our visual traffic detection system. A few months ago, we hosted FAA and EASA auditors for the Stages of Involvement 2 audit (the second of four major audits, which reviews the system requirements and design) for our visual traffic detection system. Two audits remain and although we cannot control the timeline of a regulator’s certification process, we see no substantive reason we could not complete the last two audits in the coming several months.
“It’s hard to overstate the significance of certification of the first AI-enhanced system for safety-critical applications in aviation. And frankly, it is not in line with our corporate culture to blow our own horn,” stated Gouverneur. “But as a longtime aviation engineer, I am awestruck by the work that has been done over the previous several years here at Daedalean. The contribution Daedalean engineers are making to safety and efficiency will be dramatic and long-lasting.”
No AI-based, safety-critical application has yet achieved certification in civil aviation. Industry has been waiting for years to see how regulators would approach certifying systems with AI components. Daedalean has demonstrated an approach, and if we are able to convince regulators that we know what we are doing, it will set an historic precedent.
AI in the sky is nigh
Following certification of our visual traffic detection system, we aim to grow our product portfolio to bring to market wire detection, as well as expand our visual capabilities for poor weather (including fog, rain, clouds) and nighttime operations, among others. Responding to industry demand, we will also continue to develop an ultra-low SWaP, industrial-rugged grade computing platform.
Developing such an array of applications that benefit from machine learning required our internal teams to create the tools, processes, and hardware necessary to certify them. Recognizing that this ground-breaking intellectual property could open the way for a plethora of applications making use of the techniques we’ve developed, we are bringing this IP and services to market.
“Originally, we developed new tools, processes, and hardware solely as a means to an end—to certify Daedalean's AI-enhanced systems. But our customers, who are experts in designing and developing their own sensors and systems, want access to the AI revolution. They trust Daedalean to enable them to grow their markets and segments,” stated Gouverneur. “With our expertise in certifying systems utilizing machine learning, we are uniquely qualified to become the only one-stop shop to guide developers through the certification process.”
Daedalean will be the one-stop shop for certifiable AI-enhanced automation. What does that mean?
We will offer a comprehensive product: IP and services enabling our customers to raise safety performance, reduce time-to-market, expand their product roadmap and capabilities, and capture new and growing markets. We will provide pre-qualified capabilities, including certification evidence for automation and autonomy functions, mission and task management, and perception and multisensor fusion, integration services, certifiable AI systems, data collection, analysis, and model development and verification. This new offering will enable us to catalyze the entire aerospace and aviation market with our core competencies: developing certifiable AI-enhanced software and hardware.
On the hardware side, we already announced we developed the Daedalean AI Accelerator (DAIA), which is the only inference processor board designed by a leading expert in certifiable AI for aerospace, and that DAIA will ship with the Daedalean Tensor Accelerator (DTA) firmware, which boosts performance for the convolutional neural networks Daedalean has pioneered. We look forward to bringing this indispensable hardware to market to support certification efforts for AI-enabled systems.
“It’s going to be quite a busy year from demonstrating products to pushing toward certification to offering new services,” stated Gouverneur. “Our team’s dedication to increasing safety and efficiency in aviation is set to have a visible and significant impact this year. As we eye the coming ten year anniversary of Daedalean’s founding, we are proud to deliver on the goals we set and to focus our vision toward new and transformative innovation.”
