25–29 June 2018
Hyatt Regency Atlanta, Atlanta, Georgia

Job Number 1: Detect and avoid

Posted: 1 May 2017


If engineers succeed in winning FAA approval for sensors to prevent unmanned aerial systems from crashing into buildings and other aircraft, they will be clearing the way for UAS to perform jobs ranging from pipeline monitoring to package delivery. For now, UAS developers are wrestling with the challenge of making onboard detect-and-avoid technology small and lightweight enough to fit in the fastest growing segment of the market — UAS under 55 pounds.

Contributor Debra Werner discussed detect-and-avoid technology and its implications for aircraft design with Jay Gundlach, a pioneer in UAS design, and Michael Guterres, a leader in efforts to integrate drones in national airspace. Gundlach established his own firm, FlightHouse Engineering, in 2016 to help commercial and government teams create unmanned aircraft of all sizes to perform specific missions. Guterres leads an FAA initiative to help BNSF Railway find ways to use drones to safely inspect tracks far beyond the view of their operators. Guterres also leads a multi-institution research partnership focused on integration of small drones in urban areas.

Q: What is challenging about developing detect-and-avoid systems for large and small unmanned aircraft?
Michael Guterres: Most of the larger unmanned aircraft that fly at higher altitudes are optionally piloted modified manned aircraft, military derivatives or military systems altogether. Those tend to have a lot of capacity in terms of internal volume, power and the ability to carry systems and equipment. Also, they fly in airspace that requires some type of equipment onboard. Smaller UAS fly at lower altitudes and have a different set of challenges: the necessity to avoid other aircraft, but also to negotiate ground obstacles such as buildings, cranes, trees and even people. Those small, light unmanned aircraft do not have the ability to carry or power a lot of equipment.

Q: What is happening in the large UAS category?

Guterres: There has been a pretty significant standards development effort. The RTCA [Radio Technical Commission for Aeronautics] brought together industry and government to develop a set of requirements and performance thresholds for larger aircraft, like Predator and Global Hawk, to transition to Class A airspace [above 18,000 feet]. Those standards, when published, will make it easier for operators using some of these larger systems to transition through those first 18,000 feet to Class A airspace. That transition phase from ground to 18,000 feet is a little bit riskier [than flying above 18,000 feet] because there are a lot of folks flying in that airspace that are not necessarily equipped. The UAS has a responsibility that normally would be handled in a manned aircraft by the pilot and co-pilot. When the aircraft are in Class A, they work with air traffic control. Some of the larger unmanned aircraft are able to carry TCAS [Traffic Alert and Collision Avoidance System], a system used by manned aircraft, and also onboard radars.

Q: Is it more challenging to provide detect-and-avoid capabilities for smaller unmanned aircraft?
Guterres: In some ways, yes. There are no defined performance requirements, no published standards for what a detect-and-avoid system should do around buildings, people, UAS, and to avoid manned aircraft. On the small side, I think you’ll have a combination of onboard systems with ground systems as well. For example, small UAS flying over infrastructure, pipelines or a rural railroad, where you have very little air traffic activity and low population density, may be able to use one type of detect-and-avoid solution. For UAS flying in a different type of environment, urban or suburban, close to people and buildings, the threshold for technology performance will be elevated. Then you may need different types of detect-and-avoid systems, onboard, ground or a combination of both. The technology solution may also depend on the operational concept and the risk level.

Q: In terms of onboard systems, what options are there for small UAS?
Guterres: There are quite a few that have been experimented with, from onboard radar to visual systems, using cameras and interpreting the imagery to identify objects and other aircraft, to acoustic sensors. These sensors are looking for aircraft that are not engaging with you. On the cooperative side, you have electronic communications: ADS-B [automatic dependent surveillance-broadcast] or Mode S transponders. If you have multiple aircraft using transceivers to communicate with each other then they can communicate speed, heading and altitude electronically. This makes it a little bit easier and safer, but there is no mandate for all aircraft at low altitude to have those. So, you have to contend with other aircraft that are cooperating and those who are not.

Q: How do aircraft designers integrate detect-and-avoid systems in unmanned aircraft?
Jay Gundlach: There are a few different design considerations. The primary considerations are size, weight and power. Especially for small unmanned aircraft that weigh less than 55 pounds, which are covered by the current [FAA] Part 107 regulations, you really don’t have that much capacity. So, for example, a lot of unmanned aircraft tend to have payload capacities of about 5 to 20 percent of the takeoff gross weight. For a 50-pound [23 kilograms] unmanned aircraft that’s only 2.5 to 10 pounds of payload. For a 10-pound UAV that’s half a pound to 2 pounds of payload. So a 2-pound detect-and-avoid system has a tremendous impact on those vehicles.

Other considerations are the required field of regard, all the azimuth and elevation angles that the sensor can view. Then, there’s the field of view, where the sensor can see at any given instant. For a small unmanned aircraft operating at less than 400 feet, most of the air traffic would be above it. It would not need to look below for collision risk. For low altitude flight, it might need a bump on top of the aircraft that might look like a satellite communications antenna dish that you might see on a Predator or Global Hawk. But if the unmanned aircraft is flying at higher altitudes, it might need to look below itself as well. If it is a slow-moving unmanned aircraft, things may come at it from any orientation, including from behind. A fast-moving unmanned aircraft probably wants to look more in front. All these considerations dictate what the sensors need to see.

Another consideration is what kind of sensor the unmanned aircraft is operating. Is it RF-based or an optical sensor? If the sensor has line-of-sight obstructions from the wings, tails or fuselage, that can block where it can look. For a radar-based system, it might have some nonintuitive, non-line-of-sight interactions from the aircraft, especially if it is operating at low frequencies. There are a lot of competing requirements.
Some sensors may also be transponders, like ADS-B. The designer may have to consider frequency computability with other essential functions, such as the command-and-control link and the payload downlink. To a large extent, installing a detect-and-avoid system is not that much different than trying to install other payload types.

Q: Will every unmanned aircraft model need its own detect-and-avoid solution?
Gundlach: The detect-and-avoid system integration may be airframe-specific. For example, camera-based systems will need to provide a sufficient number of sensors with the correct positioning and orientation on the airframe to provide the necessary field of regard. The most convenient locations may be blocked by elements of the airframe, which may necessitate redesign. In contrast, an aircraft that is designed to accommodate a detect-and-avoid system upfront may avoid these difficulties.

Q: Is it better to think about detect-and-avoid capabilities as you design the aircraft?
Gundlach: Correct. Otherwise you can picture some of these sensors located remotely to the primary aircraft structure in order to get the required field of regard. Or having to locate multiple sensors to avoid obstructions, for example.

Q: As you develop and test detect-and-avoid technologies are you focusing on size, weight and power?
Guterres: Size, weight and power are critical. Typically, you start with an existing UAS. The aircraft is a piece. You don’t often have the flexibility to modify the aircraft to accommodate some piece of equipment. So, system weight and power requirements are critical. That often becomes a funnel for picking options. Then you integrate it, figuring out how it works and make the best of it. However, if there is a significant enough impetus, there may be cases where the aircraft is modified to accommodate a certain system onboard.

Small unmanned aircraft systems bring a lot of advantages: low cost, easy transportability, and flexibility as far as supply chain. It has not been our observation that people are moving to bigger systems to accommodate onboard technology. There is an expectation that the onboard technology has to get smaller, lighter and use less power. The technology is evolving in that direction.

If you start with an existing aircraft, you are pretty limited in what you can do. If you can modify the aircraft, you are a little bit more able to accommodate things. Still, you quickly bump up into a load factor. If you add 10 pounds to the payload, you are most likely going to add quite a few more pounds to the aircraft itself. Staying with small aircraft is a very important thing, in general, to the operators we come across.

Gundlach: Larger detect-and-avoid systems can also make a difference in overall risk. If we were to add detect-and-avoid systems to an unmanned aircraft, then to achieve a similar level of performance in terms of payload capacity and endurance, we would need a larger aircraft. Now that new aircraft is less likely to collide with other air traffic. However, the consequences if there was an impact would be greater because it’s a heavier aircraft. Also, the risk to people on the ground may go up as well.

Generally, there is a trend toward unmanned aircraft going down in size, but depending on what the detect-and-avoid technology is, this could be a counterpressure that pushes toward larger and heavier aircraft.

Guterres: It’s a very interesting double-edged sword. An airplane can be a little heavier and much, much safer because of the technology you put on it. There is a bit of a fine balance.

Q: Is Mitre conducting research and testing specific detect-and-avoid technologies?
Guterres: Yes, for both small and large. We work on ground-based and airborne sense-and-avoid research for both military and commercial applications. Our research includes technology development, systems integration and implementation as well as standards development. Ground-based detect-and-avoid systems are generally radar installations that help with the transition through areas where there may be aircraft that are not communicating. On the civilian side, we are very involved with a program called Pathfinder, which the FAA initiated with CNN, PrecisionHawk and BNSF Railway, utilizing platform sensors and ground-based sensors as well. The FAA Pathfinder program is investigating several solutions from airport radar to small, deployable ground radar, airborne radar, low-power ADS-B, to other types of transceivers, acoustic sensors and a number of different technologies for communications as well.

Gundlach: Some of the technologies I’ve seen for collision avoidance are LIDAR [light detection and ranging] sensors that can build three-dimensional models of the world and use laser pulses to determine the range to the nearest object. The minimum size for those tends to be about 5 pounds. Smaller unmanned aircraft seem to be using either standard cameras or cameras modified to give you optical flow capabilities. Algorithms determine how quickly pixels move relative to one another to characterize the vehicle’s proximity relative to objects it is trying to avoid. Another option is stereoscopic cameras that do photogrammetry. Either multiple cameras look at the world through different perspectives concurrently, or a single moving camera views the world over time to produce a three-dimensional model, that is updated periodically, of the area it is trying to fly through. That, combined with path planning algorithms can help the aircraft avoid collisions based on the system’s capabilities and rules that have been established. The good thing about avoiding collisions with fixed terrain on the ground is that it’s in front of the vehicle, with a forward field of regard. That’s a lot easier to integrate than an aft field of regard on an aircraft. I’ve seen imagery from MIT’s Robust Robotics Group of a fixed wing unmanned aircraft flying through parking garages and performing other very demanding tasks. That technology can be made very small and low power for obstacle avoidance.

Guterres: It’s terrific for indoor flying.

Q: Not for avoiding other aircraft?
Guterres: It can be used as an aide. There were several projects in the past exploring whether this type of system could be used as a primary detect-and-avoid system with capable and powerful cameras on the nose of the aircraft. There were several research projects the Defense Department and civil government agencies funded. There are some challenges with consistency of the imagery. If you have different light conditions, if the approaching aircraft comes at different angles, it’s very hard to analyze the pixels. It is hard to do it in a way that is repeatable. As an aide, it is very capable. As a primary or solo system, it has some limitations. This, of course, could change as the technology matures.

Q: How do you see detect-and-avoid systems evolving?
Guterres: There are a couple of things that are unique and can be perceived as challenges but are also tremendous opportunities. One of them is the sheer size, the number of aircraft. This technology is somewhat in its infancy, not the pieces of the technology but the integrated technology and its integration in the airspace. Fast forward a few years you may have one order of magnitude or two orders of magnitude more of these unmanned aircraft systems flying than manned aircraft, especially the lower altitude, lower-cost systems. That brings with it scale, which enables all kinds of opportunities to reduce costs and advance the technology more quickly, but also in terms of managing these aircraft. Today the concept of small aircraft embracing some type of universal communications system among themselves may be foreign. Maybe it will not be so foreign one day. The idea of having some type of airspace management system that is more automated than it is today could be explored as a way to help reduce the risk of collisions. If you are managing fleets and flight plans systemwide, then if one aircraft is not where it’s supposed to be you can adjust the position of others. Maybe then we wouldn’t rely so much on having a silver-bullet piece of equipment on the airplane.

There is another component. As consumer electronics technology is becoming smaller, faster, better, cheaper, that directly applies to some of these small systems. Also, some of the smart cities and ground transportation automation could be leveraged. Maybe air traffic at low altitude becomes another component of smart cities.

Gundlach: Despite the proliferation of commercial unmanned aircraft operations, it certainly has a lot more potential to grow. This past summer the FAA released the Part 107 rules that govern UAS that weigh less than 55 pounds. It looks like the vast majority of UAS currently in operation weigh less than 10 pounds. That’s where we will continue to see a lot of activity. There are several key restrictions there, such as flight below 400 feet, and the pilots can’t operate beyond visual line of sight or during the nighttime without waivers. But there are so many missions that would benefit from beyond visual line of sight operations. Take for example, linear infrastructure inspection of power lines, pipelines, railways and roadways. Those operations only make sense when you can have UAS operating a very long range.
The FAA’s requirement to operate within visual line of sight is almost a recognition that perhaps the technologies for beyond visual line of sight operations, both in the onboard and ground systems, are not as mature as we would like them to be. As beyond visual line of sight technologies evolve, there are many valuable missions that would be enabled. They probably would be performed by unmanned aircraft that weigh more than 10 pounds but still less than 55 pounds. They could probably accommodate some onboard sensors, but their size, weight and power capacities are quite constrained. Detect-and-avoid systems are going to be critical for enabling acceptance from regulatory and public perspectives.

At the same time, I hope the FAA makes some accommodation for the small systems so it doesn’t limit this rapidly growing industry. There has to be a sense of proportionality. As an extreme, requiring a 5-pound sensor on anything that would be allowed to fly could really significantly hurt the UAS industry. It would drive up the system costs. The vehicles would get much larger and the barriers to entry would go up. And larger vehicles create increased severity if there was an air-to-air collision. It seems that there are definitely some cases where these technologies can open up important missions. But at the same time, there is a place for them. Perhaps for visual line of sight operations, detect-and-avoid systems would not be as necessary for people to operate safely while maintaining situational awareness. If there was a modification to Part 107 that allowed beyond visual line of sight operations with appropriate sensors, that would be tremendous and we would see a lot of important missions opening up.

Guterres: I agree the visual line of sight limitation is necessary right now and there is definitely a recognition that beyond visual line of sight will open up tremendous opportunity. The FAA started this Pathfinder program that is exploring exactly the types of challenges Jay is mentioning. It is cooperative research between the FAA and BNSF, exploring how to fly beyond visual line of sight in rural environments and low altitude. It is still a limited risk environment. We have primarily talked about the technology. Another question is how the FAA accepts those technologies. How do companies demonstrate their performance and tie that to the mission risk, which is different if you are flying over a railroad in the desert as opposed to flying over a bridge in a city. While the system may be generally doing the same thing, the level of equipment might have to be different because the risk is different.

The requirements for sense and avoid and the types of systems you have onboard or on the ground may be different depending on the level of risk you are engaging in. As Jay pointed out, the opportunities are tremendous in terms of efficiencies and things you could do that you couldn’t do before. DoD found that out many years ago when unmanned surveillance systems began providing data. Many good things came from that in the military area. In the civilian arena, you are starting to see that awakening as well. There are things that are possible that we envision today and there are things we haven’t even thought of that will reveal themselves once these systems are up there.

Q: Any other comments either of you would like to make?
Gundlach: We can take a look at the past for inspiration. About 10 years ago, Insitu put a Mode C transponder on a ScanEagle. There was no system that would be suitable at the time, so Insitu worked with Sagetech to develop what was at the time the world’s smallest Mode C transponder. This was a successful early attempt to operate a small UAS cooperatively with other air traffic. If there is enough interest and a need, systems can be miniaturized appropriately to go from manned aviation classes to small UAS. I’m confident that given enough time, funding and a supportive regulatory environment, some very capable systems can be brought down to the small UAS domain.

Guterres: A unique feature of this small UAS world is speed. The speed with which things evolve is completely different than in traditional aviation. Aviation, a mode of transportation that is mature and very safe in the United States, is now trying to embrace this very quickly developing, quickly growing new entrant. That is not without its challenges. But there may be an opportunity beyond the initial phase we are in. The technologies being developed for the unmanned systems may at some point migrate to the manned systems, making them safer. I look forward to seeing that happen because when you look at the numbers of people who have gotten licenses to fly the small UAS under the FAA rules, the number of vehicles registered is so large. I can imagine that once successful safety systems are fielded for these little craft, they will migrate to general aviation and other classes of aircraft, making manned aviation even safer.


Return to Aerospace America Home 

Return AIAA Home 

Return to 2017 AIAA AVIATION Forum Home 


FAQsContact UsPoliciesMedia