Kevin Sartori (left) and Lorenz Meier (right) Co-Founders of Auterion
What prompted you and Lorenz to start Auterion?
We have been on the same mission for a decade. What started as the Pixhawk student project at ETH Zurich in 2008 and enabled a new industry by connecting the community of developers, contributors, and companies and creating better technology the open source way. While Lorenz Meier pursued his Ph.D. at ETH Zurich, I moved to the Bay Area for my MBA in Berkeley and joined 3DR from 2015 to 2017 where I lived the transformation of the industry first hand.
2016 and 2017 were challenging years for the drone industry. Drone companies had to change business model (3DR, Airware, Skycatch), many changed their leadership teams (Yuneec, Airware, Kespry, PrecisionHawk) to help steer the company in new directions, and some are not active in the drone industry anymore (GoPro). The industry is shifting from vertically integrated to horizontally integrated; it’s a trend that we have seen in many maturing industries. It’s becoming clear that we need to work together and align with global standards and use a common infrastructure. More and more drone companies are seeing PX4 and Dronecode as a way to consolidate their efforts on a common platform. It allows them to stay on the technology curve and gain market access for their product and services to remain competitive. However, companies have needs that the open source project on its own cannot satisfy.
For this reason, we founded Auterion, a company that can address this needs and sustainably grow the ecosystem. Auterion operates on an open-source business model based on open code, contributions to the development community, professional quality assurance, and subscription-based enterprise support. We release an open source managed distribution of the Dronecode flight stack so that drone companies can leverage the community’s work and make further adaptations and improvements.
2018 will be an exciting year for open source on drones and Auterion will bring the level of professionalization to drones that Red Hat brought to Linux.
Why has open-source been so important for the drone system to date?
Open source is an R&D model and has been a critical enabler in many industries: from enterprise computing to website design, from genomics to the cloud. The same applies to the drone industry. In the beginning, there were two main reasons. First, drones are complex mechatronic systems that require in-depth knowledge in many engineering fields such as computer science, electronics, and mechanics. It is difficult for one company to hire all this expertise. An open source approach allows multiple companies to work together with researchers in academia and with the passionate developer community, making sure to cover all the knowledge gaps. Second, the open source model prevented developers from repeating the same mistakes of others and allowed the industry to innovate at a much faster speed. Cheap and reliable sensors and increased compute capacity accelerated this process, so it’s a coincidence that all this happened at the same time as the introduction of the first iPhone.
Today, in a maturing industry, there are new fundamental benefits of open source. First, open source platforms like PX4 and Dronecode align companies on a universal standard; this creates networks effects because it’s more convenient for companies to work with an ecosystem to scale the integration of their technology (Intel, ARM, NXP) and services (Airmap). Second, open source platforms provide the core infrastructure that is used to build products and services. These are things like communication protocols or the computer vision pipeline. This infrastructure is non-differentiating and it would be prohibitive for each single company to develop its own. Third, open source allows companies to stay on the technology curve and still have the freedom to customize on top. Their product or service improves as the platform progresses. All these factors make the decision to align with an open source platform a business decision more than a pure technology one.
Can you explain PX4 and Pixhawk and why they are important to readers who might be new to the space?
Both projects originated at ETH Zurich under the guidance of Lorenz Meier as efforts to enable autonomous flying. PX4 is an open source autopilot that powers any drone: multicopter, fixed wing, and VTOL. Over the years there have been adopters that used PX4 to control mobile rovers, boats, and even submarines. In Linux terms, it would be the equivalent of the Kernel.
The PX4 Autopilot is open source and free to modify under a permissive license (BSD). This fact (licensing) differentiates the project from other open source projects and is the reason for its broad commercial adoption. Today it powers products from mass scale consumer drone companies like Yuneec and Airdog and commercial and industrial products like Wingtra, Sentera, and Alti.
Pixhawk is an open hardware reference design. It explains how to create the electronics for an autopilot so that they are compatible for PX4. The design is open and was created by the development community under a creative commons license. It was successful because for the first time an integrated system included all the needed components of an autopilot (IMUs, compute, GPS) for a low cost and in a very small form factor.
Both PX4 Autopilot software and Pixhawk hardware were key drivers in the development of UAV technology for the global academic and developer community and companies worldwide.
What should new drone hardware companies be considering when it comes to leveraging existing open source projects?
As discussed before, the adoption of a specific platform, whether it is proprietary or open source, is not only a technical but mainly a business decision. The main criteria I would evaluate are: ease of use, critical integrations, market adoption, type of license, and cost.
First, to assess the ease of use of an open source platform you need to understand the quality of the documentation, the access and freedom that is provided with the APIs, what type of support is provided (if any), and whether there is vendor lock-in. These are all aspects that will make it easier and quick for your developers to get up to speed and work with the platform.
Second, integration and adoption go hand in hand. You need to assess what platform is supported by the relevant hardware vendors (sensors, chipset) and that it is integrated with the leading service providers. These platforms are often the ones that have network effects and have the most significant commercial adoption.
Third, the type of license will profoundly influence your business model and the way you can generate IP. A permissive open source license mixes the best of both worlds: It gives a company full freedom to combine open source code with its innovations without IP risks and license complexities. It also enables a company to leverage a mature codebase.
Finally, you need to consider costs. Open source platforms are free to use, but this doesn’t mean they have no cost. You need to invest in the platform to bring your development team up to speed, continuously maintain the platform and adapt to the changes in the industry or regulations. This support is the core value we are providing with Auterion.
What technologies are you most excited about being integrated into drones in the coming years?
There is nothing more fundamental in any robotic system than the ability to perceive, communicate with, and navigate the world around it safely.
First, we first need better detection and avoidance technologies at a fraction of its current cost.
Second is airspace integration. The initial step in technology was demonstrated in last year’s U-space demo in Switzerland with PX4, Airmap, Skyguide and the Swiss Federal Office of Civil Aviation flying a the Intel Aero drone. The next step for the industry will be to work together with regulators to understand use cases and define rules to manage air traffic of manned and autonomous systems safely.
Third, like every IoT device connected to infrastructure, drones need to be reliable and secure. We saw last year with the ban on DJI products by the US Department of Defense that cybersecurity is a concern and not yet a solved problem.
Finally, APIs and integrations are critical to bringing drones online. Together with our partners, we are contributing heavily to the addition of vehicle-to-vehicle links. We push the boundaries of computer vision and enable drones to fly safely, avoid obstacles and seamlessly communicate with each other and with the UTM system.
Where would you like to see drone hardware improve in the future?
It’s simple: hardware needs to become cheaper and better. Drone hardware has not yet reached the maturity that smartphone hardware has. Today, drones still resemble cell phones in their infancy, and hardware constraints limit the user experience.
To reach the next level of adoption, drone hardware needs more visual computing power and better sensors at lower cost. A drone built to deliver highest-end user experience will still cost a few thousand dollars — cost needs to drop by order of magnitude to provide a truly consumer-ready user experience. Luckily, the automotive industry is solving the same autonomy problems and will be the primary driver in reducing cost because it offers much larger scale and returns on investment than the drone industry. We are working with leading technology providers in the automotive industry to cross-utilize technologies on drones.
Parting thoughts?
If your business, technology, or services would benefit from having access to an ecosystem, I encourage you to join Dronecode. Members directly contribute to the Dronecode platform development, influence priorities, while enabling their products and services to work with the platform.