This series of articles will discuss the benefits of AI in the drive for autonomy and BVLOS operations, initially in drones, and then in the broader commercial aviation environment. It will explore some of the benefits that can be gained, the challenges that are faced, and propose some solutions to solve the most persistent and challenging problems. It will further explore the role of drones in a broader ecosystem, and how essential it is for machines and people to work as coherent teams, identifying the challenges and opportunities that this presents.
Negative Ghostrider, the pattern is full – the requirement for autonomy and Human/Machine teaming in future airspace.
Today air traffic management is heavily dependent on humans. This dependence has, on countless occasions, saved lives, averted disasters and generally helped keep the skies moving freely.
Our Air Traffic Controllers have a tough job, spinning multiple plates at any given time, working under high pressure and managing risks dynamically. Communication, complex problem solving and analytical skills are critical to success. The air traffic controllers are a core element of the high safety levels that exist in today’s airspace. ATCs are broadly split into three roles – Aerodrome Controllers, Area Controllers and Approach Controllers, with different areas of responsibility and control over airports and airspace.
As aviation continues inexorably to move to a hybrid autonomous model the needs and demands on air traffic controllers must and will change to match this model. For example, today, voice communication is still the primary and critical method of communication between aircraft in the sky and the controllers on the ground. Indeed, remotely piloted aircraft operating beyond visual line of sight (BVLOS) are required to have voice communications between the pilot and ATC.
Autonomous aircraft do not have pilots. It seems like an obvious statement, but this means that they also have no voice, so how do they communicate with ATC? Or with other aircraft?
Data communications will continue to grow in significance over the coming decades, slowly at first, but with increasing pace. Linked to this will be the ability for autonomous aircraft to share airspace with manned aircraft, which poses a number of questions:
At a system level, how do you manage autonomous and manned flights together?
How can an air traffic controller manage autonomous flights within controlled airspace?
What is the role of the air traffic controller in the future?
Managing Autonomous and Manned Flights Together
This is possibly the single biggest challenge facing the autonomous aviation industry today. It’s no coincidence that manufacturers of large autonomous aircraft recognise and are designing for the need for pilots in the first instance. The industry as a whole does not have a solution, or a roadmap today. A change of such significance will undoubtedly add risk, and risk is the enemy of progress in aviation – rightly so, but it also cannot block it entirely.
Today, risks are mitigated by adding smart, well-trained, disciplined people into the critical parts of the system (pilots, ATC, etc). Tomorrow the network will need to be managed as a human/machine team, with smart people and smart machines working together in harmony. The people will remain critical, but their roles and inputs are likely to change considerably from today.
As the number of autonomous vehicles increases it simply won’t be practical for them to be managed individually by humans in the way aircraft are today. Instead they will need to be managed in a manner that works for all actors in the chain. This will also mean that today’s methods (such as voice communication) will no longer be appropriate. Integrated human/machine teaming platforms and systems will be required. These must be able to manage complex, high-volume, high-pace analysis of situations, providing easy to interpret options modelled in synthetic environments to the humans in the loop - to allow the final decisions (and therefore responsibility) to rest with them when any extraordinary events occur.
Today it is inconceivable that the network could run fully autonomously, with humans excluded from the loop, I hope that some of us may just about see it in our lifetimes.
It’s important to recognise that the general mobility environment is likely to become more complex, with a growth in multi-modal journeys and “last-mile” autonomy playing a significant role in the movement of goods and people in the future.
Our experience developing human/machine teaming solutions can provide a unique insight into the challenges to be overcome, and we welcome the chance to share this experience with industry.
Managing Autonomous Flights in Controlled Airspace
This is a huge challenge for the industry. How do you manage autonomous flights in a hybrid airspace? Do they get priority? What happens if there is a problem with an autonomous air vehicle? How do you secure the vehicle from being controlled by those with malicious intent?
What is clear is that the current method of each aircraft being assigned a specific human air traffic controller is not scalable or practical. Autonomy is a prerequisite for the growth of the aviation industry as it enters this next phase of evolution, as is the assignment of specific areas within the skies for safer autonomous flights.
The real question in this context is actually “How do you certify AI for use as the primary flight control system on autonomous aircraft?”. Once this is established, one can ask the follow-up questions like “How does a certified AI system change the way we manage air traffic in the future?”. What does UTM (Unmanned Traffic Management) look like in real terms?
As with any system design, understanding is key to success. Mapping out the workflows, identifying areas of risk, building mitigations in, testing, validating are all essential. When automating systems one regularly discovers that human intelligence regularly overcomes significant process or systems gaps – think about how often you start a new job and are told “we do things this way because...” – how often is that because the underlying process is incomplete or underperforms?
I regularly hear from people in the AI world about all the great things that their technology can do, but in aviation it is more critical to understand what it can’t. I always remember talking to the British Airline Pilots Association about autonomy, their entire mindset was based on the “what happens when things go wrong” approach. It was clear to me that they saw their role not as someone who flies the plane, but more as someone who stops it crashing. Like they said to me at the time “planes can already fly themselves, that’s not what we are needed for”. The same approach needs to be taken at network level. What mindset should we champion as an industry to ensure we are answering the right questions?
What is the Role of the Air Traffic Controller in the Future?
These questions are massive and complex and cannot be answered today, other than to say that the approach requires collaboration, communication and understanding from all actors involved. At the time of writing, the expectation is that rather than monitoring each aircraft, the air traffic controllers of the future will monitor areas of the skies and identify specific challenges/areas of interest.
How do we Make Progress?
This is the fundamental question that leads us forward. As we’ve established above, the path to autonomy is complex, challenging and carries significant risk that needs to be retired. For us to make progress we need to work as an industry, embrace the benefits of human-machine teaming, taking appropriately-sized steps towards a common, clear goal. Risks will have to be mitigated at each step of the way, and it is very unlikely that any one person, company, governing body or even government will have all of the answers. Communication and collaboration, in all senses, is key.
We at Archangel Imaging are dedicated to this journey and would love to travel it with like-minded people. Get in touch and start a conversation below:
Comments