Blog

Hold the future of autonomy with Daedalean’s evaluation kit

Real autonomous flights are not here yet. We are working on this intensely, but it takes years to develop and certify critical safety instruments. However, this does not mean that you must sit and wait patiently until we finish. In fact, the sooner you start preparing for it, the quicker you can benefit from it yourself, as well as influence the product.

Some may assume that when the technology finally arrives, they simply need to purchase it, install it, and press the button to switch it on: "Yay, it seems to work! I can fire half of my pilots or even all of them on the spot, and the smart AI computer thingy will do all the flying tirelessly and error-free, and it will never ask for annoying stuff like salaries, shorter shifts, insurance, or paid medical leave." But no—it won't be like this.

To paraphrase William Gibson, the future will soon be here, but it will not be evenly distributed. To reap the fruit of the autonomous revolution, you need to do your part of the homework. The preparation is a long, exhausting, and incremental process, and we suggest you start it as soon as possible.

Consider the case of Amazon delivery drones. With great fanfare, the company announced an experimental drone delivery service back in 2013. It's been almost nine years. But do we routinely receive our Amazon packages containing books, cat food or toothpaste from drones dropping them on our doorsteps? Well, nope. Aside from a few exceptions and local trials, it's still old-fashioned human couriers who do that.

Source: Mike Licht, on Flickr under Creative Commons license

For all these years, Amazon has been honing and testing the technology, developing operating guidelines and safety measures, and working with regulators, authorities, and more. The first successful testing delivery was completed in 2016. In 2020, the FAA granted approval—a Part 135 Air Carrier Certificate, meaning that Amazon Prime Air was now designated an "air carrier." This meant that it was allowed to begin its first commercial deliveries in the U.S. under a trial program. Still, only a trial program! When will drones finally be routinely delivering my grocery orders? Amazon's website does not provide any particular launch date, saying, "The only solution worth launching is one that is safe and reliable. We are actively flying and testing, and it will take time and more hard work before our operations are ready to scale."

It took the company seven years to convince the government that they were ready to operate in the highly regulated aviation sector, and now, two years later, we can only assume that Amazon must still clear some imposing regulatory and technical hurdles. And that's just small delivery drones, mind you—it's not like they will transport people.

Another case in point would be the history of ETOPS—ETOPS standing for Extended-range Twin-engine Operations Performance Standards—which establish your allowable flight length from the nearest airport suitable for an emergency landing if one of your two engines fails. This part of the flight rules was introduced in relation to the first aircraft able to perform transatlantic flights. In the 1950s, this timeframe was 60 minutes. With advances in technology and the development of jet engines, this became 90, and then 120 minutes. Piston engines of the past were unreliable, but that has changed: the current approval standard for 180-minute ETOPS is 0.02 shutdowns per 1,000 hours of engine operation.

The Airbus A300B4 became the first ETOPS-compliant aircraft, in 1977
Source: Mike Freer, under GFDL 1.2 on Wikimedia

To quote Simple Flying, "In 1985, the first ETOPS certification rating was given: ETOPS 120 minutes. ETOPS 120 became the standard but this gave way to ETOPS 180. Achieving this increased rating was only possible after a year of trouble-free 120-minute ETOPS experience." It is a long and rigorous process to win authorities' trust in the technology step by step, with each subsequent step becoming possible "only after a year of trouble-free experience" with the previous one. It's similar to parents allowing their growing children to come home later and later until they feel comfortable letting them sleep somewhere overnight.

The same process will, without a doubt, be taking place with approaching autonomy in the coming years. Getting approval is a multi-step process. When the equipment is certified and launched to the market, aircraft operators will still need to convince their own country's aviation regulators of their ability to use our avionics to conduct AI-assisted and then increasingly autonomous flights.

When the technology is ripe, an operator with extensive experience in AI-assisted flights may be awarded the operational approval immediately, while others may need to accumulate more experience. And as we've seen in the case of Amazon, this may not happen quickly.

Become an Early Adopter

We at Daedalean have reached the stage of development where we are ready to demonstrate our computer vision capabilities and collect flight data. You can finally board this train and start recording your own history of rolling in the autonomous flight. We offer an evaluation kit—Eval Kit for short—for assessing the performance of computer vision technologies in a real-world environment. This allows you to explore their capabilities as a source of redundant flight information, get the full idea of what the eventual product will be capable of, and align your development roadmap accordingly. By the time the actual product is certified and launched to the market, you will already have a long history of testing it, trying to break it in all possible ways, assessing the risks, marrying it with your equipment—and thus will have strong arguments for getting all necessary operational approvals, flight clearances, and certifications.


The Eval Kit is not yet a fully-functional autonomy set. Currently, it is just a seed of our future product—but this allows you to be involved in its development and influence it by setting your requirements and providing feedback. Along the way planned in our roadmap, the system will acquire night vision and non-vision sensors, become capable at degraded visual and full IMC environments, embrace the existing means of collision avoidance with non-cooperative traffic, and on top of that, will be able to communicate with traffic control services, other aircraft, and ground control for the operator. But the hard parts come first: we are now gathering the data for the technologies that did not exist before, i.e., computer vision-based situational awareness.

This includes three essential functions for rotorcraft, eVTOL, and fixed-wing aircraft:

●     Visual Positioning and Navigation
●     Traffic Detection
●     Visual Landing Guidance

Basically, the Eval Kit consists of two cameras paired with a processing box and a tablet computer as a primary UI to observe the output in real time. The installation is not that easy, though (duct tape won't be enough): the cameras must be placed outside the cabin in a fixed manner; all the cables for power and data should be laid down, and the power supply provided. You also must ensure that the (temporary) installation is approved and safe to fly.

The Eval Kit is ready to use. The output of the visual system and numeric data can be viewed on the included tablet's display during the flight. A pre-installed app connects it to the computing box that shows you real-time visual output and numeric data. Also, the Eval Kit records raw and benchmark data for extensive post-flight analysis.

And for those who are willing to delve deeper and make an engineering effort, the Eval Kit is suitable to integrate with your flight control instruments at any level you like. Your avionics, flight control, or autopilot can be set to receive data from the Eval Kit in real time. See details in the Eval Kit brochure. Or contact us to discuss your use case and how we can collaborate.
2022-01-24 12:53 Blog