Your questions answered: ASTRAEA autonomous planes

Last month, a team of engineers announced they had flown a converted passenger plane 500 miles without the aid of a human operator. They answer your questions on on how they did it.

The age of unmanned aircraft is upon us — and this no longer means small purpose-built drones and remotely operated miniature vehicles. Last month, a team of British engineers announced that an adapted conventional Jetstream passenger plane had flown 500 miles in UK airspace without the aid of a human operator.

ASTRAEA, the UK’s cross-industry project to develop the technology and regulations for using unmanned aerial vehicles (UAVs) or systems (UAS) in civil airspace, has brought together the country’s top experts in autonomous flight to achieve this goal. We put some questions to a few of those experts to find out how they did it.

Our expert panel includes:

  • Lambert Dopping-Heppenstal, programme director of ASTRAEA and engineering director of systems and strategy at BAE Systems;
  • Nick Miller, vice-chairman of ASTRAEA and business director of UAV systems at Thales UK, which developed the project’s collision avoidance;
  • Gary Clayton, head of R&T for Cassidian UK, which was responsible for ASTRAEA’s communications architecture;
  • Dr Andrew Lucas, managing director of small British firm AOS, which created autonomous decision-making software for ASTRAEA.

What have been the main achievements of ASTRAEA?

Lambert Dopping-Heppenstal, BAE Systems

LDH: The ASTRAEA programme was established to understand what is required from both a technology and regulatory point of view to safely allow the introduction of unmanned aircraft into our airspace. The programme has successfully demonstrated the key enabling technologies to achieve this and has made significant progress in understanding the regulatory framework that will be required for the safe introduction of unmanned aircraft into all classes of airspace without any onerous restrictions.

NM: Thales has been developing sense-and-avoid, probably one of the golden nuggets for airspace access. We can now detect as well as a human eye can: the object of sense-and-avoid systems is to be equivalent. We can detect systems, do automatic collision-avoidance manoeuvres and pass the information to a pilot on the ground.

How much of an impact do you think ASTRAEA will have, both on the technology of autonomous flight and on the public acceptance of it?

NM: Technology-wise we’ve made significant steps because we haven’t just made equipment, we’ve actually analysed the design requirements and worked out how a UAV system and its subsystems work together. Some people have made equipment that will go into UAV systems or back into manned assets. We have addressed a little of the public perception issue but I think that will be the next stage when we come to the users. The clear message we have for civil UAVs is that the public will accept them when they save lives or help the public. There’s a perception at the moment that UAVs or “drones” are used for war, and we’re not on that planet at all.

GC: It is important to realise that ASTRAEA is a research programme and is therefore a long way from commercial integration. The products and services that will eventually be delivered will impact both manned and unmanned aviation, providing increased pilot safety aids, assistance and communications. There is a perception that the larger systems would be used for urban surveillance when in reality this will not be the case and they are more likely to be used for tasks such as oceanic search and rescue. The use of unmanned systems routinely in non-controversial applications will build public acceptance and allay public fears that are being fostered in some areas of the press.

The ASTRAEA project brought together a number of communications, sense-and-avoid and decision-making technology.

What has been the most challenging technological aspect of the project and how have you addressed it?

LDH: All the basic technologies exist. The main challenge is the application and integration of these technologies in a manner that is certifiable. An illustration of this sub-system integration has been a series of recent demonstrations including:

  • two small, unmanned aircraft in a simulated search and rescue (SAR) mission;
  • a team of specially equipped [road] vehicles replicating the demands of a secure and robust communications network while driving through remote and mountainous Welsh countryside;
  • a robotics rig demonstrating automatic in-flight refuelling that could allow unmanned aircraft to operate for extended periods of time, e.g. while undertaking SAR operations far out at sea;
  • an intelligent and integrated power-systems rig demonstrating the autonomous operation of an unmanned aircraft’s propulsion and electrical system;
  • a surrogate UAV flying through shared UK airspace while under the control of a ground-based pilot and control of National Air Traffic Services (NATS), demonstrating transparency.

NM: The technical challenge for sense-and-avoid has been the algorithms, getting them to look at separating predictive collision avoidance and last-ditch autonomy without the human in the loop. So if a UAV detects a potential collision and advises a reroute but for some reason the pilot doesn’t accept it then the UAV has to work on its own and come to last-ditch collision avoidance and make its own moves. There are actually lots of different types of algorithms for whichever part of the scenario you’re in and it’s actually breaking those down that’s the clever bit.

Gary Clayton, Cassidian

GC: Cassidian developed an innovative ad-hoc secure mobile IP networking solution that is capable of transforming the communications capabilities of unmanned aircraft. When there are multiple aircraft using multiple routes you are unable to use satellite communications because the latency affects the work of Air Traffic Control (ATC). Mobile IP Node can create a secure terrestrial network between multiple air platforms and any other equipped platform — be it air, land or sea-based — on an ad-hoc basis. It can also manage the transition from high to low workload automatically with no user interaction, switching back to satellite communication where lone operation does not pose an ATC danger and a terrestrial network cannot be formed. It also negates the need for a pre-planned communications infrastructure based on a pre-planned flight plan. 

AL: AOS has developed reasoning software for a number of years — e.g. in defence and oil industry applications — but ASTRAEA has led us to develop a new product that is amenable to safety assurance validation and verification. We have developed an approach to reasoning for detect-and-avoid that is not just reactively acting to avoid a collision, but proactively planning the route as the aircraft makes its way through airspace occupied by a wide range of users, from parachutists to airliners.

Cassdian’s communication technology was tested in a fleet of Minis in the Welsh hills.

What have you been able to achieve in the area of automated air-to-air refuelling and could the work be applied to piloted craft?

LDH: Full-scale ground demonstrations using sophisticated robots have demonstrated the viability of ‘autonomous’ air-to-air refuelling. This development, like almost everything else that we have been doing, can benefit manned aircraft in enhancing capability and further improving safety. [Note: Unmanned aircraft are also “piloted” — there will always be a ground-based pilot responsible for the operation of the aircraft.]

NM: I’m sure that this technology and others will all help manned assets in the future. But it’s a fine balance because you can’t take man out of the loop and UAVs won’t take over from manned assets: they’ll work together. One clever thing Thales has been thinking about is that a sense-and-avoid system is trying to keep aircraft apart and when you have air-to-air refuelling you’re trying to bring them together. You’ve got to think of how not just to detect something in front of you but the whole picture.

What do you see as the future of the programme?

We now face the “valley of death”, i.e. we’re still at too early a point for individual companies to invest in commercial UAS.

Andrew Lucas, AOS

LDH: We have reached about two thirds of the way along the maturity journey to bring unmanned aircraft into our airspace without undue restrictions. The next steps are to continue our engagement with the regulatory authorities in the UK, Europe and beyond, building on the virtual certification work, and to validate the technical capability through further, integrated demonstrations if appropriate funding and national support can be arranged. More than 100 UK companies and universities have been involved in the programme to date and we would hope to continue with this broad engagement.

GC: The consortium feels [certification development] should continue but public-sector financial support is vital for the programme not to be fragmented and to develop on the achievements made to date. This funding, although proving difficult to secure, is vital in ensuring that another UK leading position in new technology is not wasted as so many have been in other areas in the past.

AL: We now face the “valley of death”, i.e. we’re still at too early a point for individual companies to invest in commercial UAS. The market is still only prospective and we don’t seem close enough to a product for early adopters to do more than watch on. We need another round of government support to bring the technology to where a company could see a product, and the CAA see how it could be safely certified for operation in civil airspace.

In what sectors and for what uses do you think UAVs will first begin to operate in civil airspace? Who will the “trailblazers” be that set examples for the rest?

GC: As can be seen from the blossoming small UAS sector, many of the applications are new and were not only done differently but in many cases not done at all. It is expected that the main applications for Beyond Line Of Sight (BLOS) platforms will also be developed for things that are new. There is a traditional view that they will be applicable to ‘dull, dirty and dangerous’ activities and this is certainly true but this new field of aviation will develop new applications that were not viable previously.

NM: Already we’re starting to see UAVs used for farming and environmental monitoring. There’ll be surveillance and search and rescue, monitoring flooding and forest fires, crop spraying, pipeline protection — all the repetitive asset-management missions.

Andrew Lucas, AOS

AL: The first “trailblazer” is likely to be the US Department of Defense. Now that withdrawal from Afghanistan is well under way, there will be a large number of “drones” based in the US. The second will be the deployment of key sub-systems developed for the UAS market in manned aircraft. Here the system is not taking over from the human, but assisting in maintaining situational awareness and helping to select the safest action to take. The first commercial uses of UAS in unrestricted airspace, I believe, will be: freight; inspection of key infrastructure, such as power lines or oil pipelines; and agricultural surveillance, e.g., of livestock or crops. I have carefully avoided the popular early adopter suggestions of search and rescue, fisheries etc., as these are government services and in the current economic climate governments are unlikely to be first adopters.

In what sectors and for what uses do you think UAVs will first begin to operate in civil airspace? Who will the “trailblazers” be that set examples for the rest?

LDH: I see unmanned aircraft as a new branch of aviation under-taking operations that are difficult, impossible or too dangerous to undertake with manned aircraft. There is little point in moving the pilot of a passenger aircraft to the ground, although, in time or for freight operations, single on-board crew may be attractive.

GC: I don’t see pilotless passenger aircraft on the horizon at all. However, the technologies developed for unmanned aircraft have the potential to dramatically increase the safety and cost-effectiveness of manned aviation with increased, ever vigilant, pilot aids. Large unmanned aircraft and manned aircraft should be seen as complementary technologies sharing airspace to conduct their distinctly individual tasks.

NM: It’s possible but it’s probably long-term. When will you get full [pilotless] passenger planes? It’s difficult to say. Nowadays the planes pretty much fly themselves anyway — we’re almost there technology wise but it’s the legal and regulatory aspects. People may not want it.

AL: I think that it will happen. However, it will be in four steps. The first will be autonomous small freight, inspection and agricultural applications. Second, autonomous technologies will be introduced into passenger aircraft cockpits to improve crew situational awareness, decision-making and monitoring. The third will be single-pilot operations of passenger aircraft, with the second pilot being a virtual copilot. I’d suggest 2020 as a target date for trial operations. Finally, autonomous passenger aircraft will appear. But by then people will be already in driverless cars travelling up the M1 and it won’t seem such a step as it is now.

ASTRAEA could lead to improved technology for manned passenger planes.

What would need to change to make this a reality?

AL: Demonstration that the technologies work reliably, and miniaturisation of some, such as sense-and-avoid sensors, and final agreement on the regulatory standards. ASTRAEA has contributed to this journey, but we need international agreement and finalisation of some aspects, such as agreed software integrity levels for some functions. Public acceptance will be straightforward if we are open and consultative, and problematic if the industry keeps to itself and then suddenly surprises the community. Our biggest asset is the young — those under 20 who are relaxed about technology and do not see a problem with computers being integral with every aspect of their lives.