Making flying safe: The evolution of DAA in unmanned drones

Iris Automation identity

Jon Damush, CEO of Iris Automation, explains how DAA systems will be key to helping aviation regulators integrate UAS’s into normal airspace

These are exciting times in the UAS space. Momentum is growing, civil aviation bodies are getting more comfortable with the necessary regulation required across the industry and of course the commercial opportunities are expanding.

These include infrastructure inspection, mining, mapping, agriculture, emergency response, and package delivery.

Story continues below
Advertisement

But work still remains to ensure we remove the risk of mid-air collisions. Closely related to that is how do we scale commercial drone operations by reducing the dependence on dedicated pilots, sometimes over one per drone? Fixing these two obstacles are the next major breakthroughs to unlocking the commercial drone space.

In the mainstream world of aviation, there are multiple checks to prevent mid-air collisions and an increase in automated tasks – and auto-pilot.

We employ a system of systems and multiple layers of mitigation to avoid mid-air collisions, from the flight plan – the altitudes chosen, the route plotted, the airspace transited all contribute to avoiding other traffic and terrestrial and environmental hazards – to a variety of tactical methods to avoid other aircraft, including talking to air traffic control, using ADS-B, ground-based radars, and transponder equipped aircraft.

Finally, we come to the last line of defense — the human pilot. It’s essentially a ‘stack’ of mitigation layers to filter out the collision risk. But no system is perfect, which is why FAR part 91.113(b) states “The pilot….shall see and avoid other aircraft.”

So when there is no ‘pilot’ on board we have effectively lost our last line of defense. The good news is that we have the ability through regulation, technology, and collaboration with each other to have all drones be considered ‘cooperative,’ reporting where they are and what their intentions are so that other airspace users are aware of where the drone is operating and what it plans to do.

Auto-pilot for drones

Systems like UTM (Unmanned Traffic Management) offer great promise to help the low altitude drone community avoid mid-air collisions due to the ‘cooperative’ nature of the drones.

Our main challenge comes from the desire to push flight plans outside the boundaries of the current regulations, such as Part 107 in the US. This allows for commercial drone operations below 400 feet above the ground, within visual line of sight, and a host of other restrictions. But with use cases like long-linear infrastructure inspections, large area mapping, search and rescue, logistics, and passenger transit you begin to understand why they must fly ‘beyond the visual line of sight’ (BVLOS).

Though talking of BVLOS, I’m not a fan of the acronym as it implies the desire is always to fly farther than someone can see. The bigger economic value is that solving the BVLOS problem inverts the critical operator-to-aircraft ratio. For drones, right now that is one (or more) to one. In many cases, operating a single drone requires more than one human operator. We will see the rapid adoption of drone technology not because they can fly far, but because one person can fly multiple drones. This is where the economic value lies with robotic aircraft.

So where are we today?

To date, we have not had an equivalent system to the pilot on drones that can provide the same ‘last line of defense.’ This is where Iris Automation is focusing our energy. We developed a system called Casia that uses a combination of computer vision, artificial intelligence, and machine learning to detect, classify, and compute the relative position of an aircraft in flight. 

We’ve been collaborating with partners and regulators to explore how our system can be used in conjunction with other technologies and concepts of operations to mitigate the mid-air collision risk as well, if not better than we do today in piloted aircraft. The exciting part of our approach is that by using computer vision, we are able to use relatively low-cost, readily available, and small off-the-shelf hardware components.

This helps keep our size, weight and power (SWAP) low enough to be placed on-board the aircraft without negatively impacting that aircraft’s ability to perform its mission and carry its revenue-generating payloads. Improvements in performance in the future will be provided through software updates, including constantly improving our machine learning database, allowing for use in more diverse environments and detecting and classifying an ever-increasing number of aircraft and other obstacles.

We’ve already achieved waiver approvals in Alaska, Kansas, South Africa, and Canada for BVLOS flight trials, which are an important stepping stone and ones that led us into the FAA’s BEYOND program to advance more complex UAS operations in the National Airspace System.

This year we hit some major milestones around regulatory testing. Transport Canada issued the second Special Flight Operations Certificate (SFOC) for BVLOS flights in uncontrolled airspace utilizing infrastructure masking and Iris Automation’s onboard detect-and-avoid (DAA) solution to MVT Geo-solutions.

Under this SFOC we will partner to conduct commercial missions over linear power lines in Alma, Quebec. These flights will mark the partnership’s first BVLOS flights outside of the CED Alma test range that leverage onboard DAA for air risk mitigation and does not require ground-based visual observers or radar.

In May 2018, the City of Reno in Nevada was selected by the United States Department of Transportation (DOT) as one of nine state, local and tribal governments to participate in the FAA UAS IPP. The UAS IPP is working with the nine public-private partnerships to implement and study specific drone applications across the United States in an effort to advance the safe integration of drones into the nation’s airspace.

We partnered with the City’s Fire Department to test equipping it with drones to conduct life-saving river search and rescue operations. The River Search and Rescue program will test the safety and capability of using drones during river rescue missions in an effort to improve response times and reduce exposure of both first responders and victims to dangerous conditions during river rescue operations.

Are we as good as humans yet? Not quite, but we are getting there fast. Plus an automated system doesn’t get distracted, tired or bored doing its job. It can see all the airspace around it and is not occluded by cockpit structure, passengers, or cargo. This makes it a critical part of the system of systems that includes UTM, radar, ADS-B, and other strategic mitigations.

Analogous to the human pilot, we are in a great position to provide the ‘last line of defense’ for collision avoidance to the commercial drone community. We see a time when an onboard DAA system like Casia will be the key to transitioning aviation regulation to accept full integration of unmanned systems with piloted aircraft.

It sounds simple, but the goal is to make flying safe.

Tags : AutomationIris Automation
Joe Peskett

The author Joe Peskett

Leave a Response