Man vs Machine: How far will autonomous UAVs take us?

Despite, or perhaps rather alongside, such apprehension, many visions of the future include the use of new aerial transport solutions that could mark the end of bumper-to-bumper commutes on highways and ever-crowded public transportation systems, particularly as cities continue to grow ever-larger air taxis.

It’s a step that players around the globe have already been working towards for decades, with billions of dollars having already been invested into research and development, across new tech-driven start-ups such as Archer and Lilium, and existing transportation giants such as Hyundai and Airbus. With the first air taxis due to hit the consumer market in 2025, the industry is only set to continue to grow exponentially in the coming decades, with current estimates suggesting that the air taxi market could be worth some $1.5 trillion by 2040.

Embraer offshoot Eve Urban Air Mobility is one of a number of major players heavily investing in the research and development of futuristic air taxis and other air mobility solutions, between developing its own electric vertical take-off and landing vehicles (eVTOLs) and scoping out realistic ways to integrate the air taxi systems of tomorrow into today’s cities and airspaces. According to Eve vice-president of business development for Asia-Pacific David Rottblatt, urban air mobility solutions aren’t really as futuristic as one might think, and actually grow upon a concept that has been “incubating ever since the 40s and 50s”. He explains that companies have been supporting commuter and tourism transport via traditional aircraft for decades, pointing to the establishment of Los Angeles Airways in 1947 and New York Airways in the 1950s. “The procedures and the infrastructure that we’ve developed to support this conventional version of what we now call urban mobility, have been very successful at getting this industry to continue to mature.”

However, Rottblatt says that the key difference in the 21st century air taxi development sector is the available economies of scale. Firstly, New Age air taxis will almost certainly be powered by electricity, as opposed to traditional jet fuels. These battery cells require many magnitudes, less upkeep and maintenance than jet engines, and can be charged via solar or wind power, making electric air taxis not only more environmentally sustainable, but also more economically viable for operators. An eVTOL’s ability to land and take-off vertically as opposed to requiring a runway also means that less space and landing infrastructure is required, which similarly cuts down on costs to establish and maintain the sector.

UAVs take the lead

Importantly, according to Rottblatt, the factor that will make the biggest difference to the next-generation urban air mobility market in the longer term – that will drive costs down for the consumer to a level that makes catching an air taxi not only a convenient but an affordable alternative to catching an Uber or bus to the airport, for example – is automation.

“I think I speak for the majority of industry, when I say that we’re excited to evolve from conventional piloted aircraft to remotely piloted or more autonomous forms of flight, as soon as possible,” Rottblatt says. “This is how we get to an increasingly affordable price point for our future passengers,” he said. “When technology is ready to manage much of the responsibilities that a pilot has done in the past, that means that you no longer have to pay for that pilot in your ticket cost. This is how we get to scalability.

“It’s a concept that I’m sure many readers will cringe at, at best, or perhaps instill a feeling a horror. The idea raises a lot of questions, that I’ve endeavoured to answer.”

Can one actually ensure the level of safety and reliability required in aviation without a pilot onboard? Are pilotless air taxis a plausible solution to growing traffic problems in sprawling urban environments? Will regulators and the flying public alike be capable of overcoming their natural skepticism, and trust an aircraft without a pilot with their life?

According to Dr Matt Ebbatson, a senior lecturer on aviation at Swinburne University, the answers to these questions, generally, are yes, yes and … maybe.

“Aviation is a heavily regulated industry that prioritises safety and reliability above all else. So the ability to build a system that provides that same level of reliability that we’ve come to expect from aviation, that is safe enough to operate in close proximity to large collections of people and preserve their way of life. And to not only do this but to be perceived by the public as doing this to a sufficient extent, I think that’s going to be the real challenge,” Dr Ebbatson says. And yet, if we are ever going to see a community that utilises and benefits from air taxis, it will be a necessary hurdle to cross.

For Dr Ebbatson, the importance of AI, machine learning and unmanned vehicle operations in the future’s air taxi industry cannot be understated, and not just for the sake of the system’s economic viability. “In fact, I would take it one step further and say it won’t be just important, it will be essential to the creation and sustainment of mature air taxi systems,” he says. “It will be the only way that such a system can operate.” 

For an air mobility system to grow to engage dozens, if not hundreds, of airborne vehicles at once, the system will ultimately become far too complicated for humans to coordinate and facilitate effectively and safely. Highlighting this fact, in 2020 Airservices Australia teamed up with Eve to produce an initial concept of operations (CONOPS) for the flying taxi market in Melbourne, with the intention of extrapolating that data to use across other markets. “What they found in that collaboration is that data becomes central to the smooth operation of that system. It needs a backbone of information which coordinates the whole system and lets the vehicles talk to each other and talk to the air traffic control system,” Dr Ebbatson says.

He reveals that during one of those proof-of-concept tests, Eve and Airservices utilised three standard-sized drones, far smaller than actual air taxis, and flew them along a makeshift corridor, to simulate the flight of an air taxi connecting Geelong to Melbourne Airport. They utilised existing air traffic controllers and tasked them with attempting to coordinate the drones as they anticipate they would, had this been a duly fledged system.

“Even with just those three drones in that simulated system, that trial saturated the air traffic control system,” Dr Ebbatson says. The controllers couldn’t cope with the extra workload because of the different rules which they have to apply in order to separate these things from valuable assets like airliners, and to allow them safe passage. So the current system is just not going to work. There is going to have to be a huge change. And that is going to rely on handing getting the human out of the system of low-level tactical decision making and letting AI take over that role.”

“The current system is just not going to work. There is going to have to be a huge change.”

“This is where technology can aid, in a way that it has been doing in aviation for decades already, albeit to a different extent,” Dr Ebbatson adds. “When we go from having tens of aircraft in the sky to having hundreds – if not thousands – then we’re suddenly in a situation where it’s going to be beyond the human’s capacity to predict how the network will respond, and how best to optimise it and control it. So much of the decision-making element of the systems operations is going to need to be automated and supported by AI technologies.”

For Dr Ebbatson, creating the technology required to see a mature and largely autonomous air taxi system in operation will “not be insurmountable”. However, as is the key issue when discussing automated aviation systems in general, something that may in fact be insurmountable is getting the public to support and trust completely pilotless UAVs in the longer term. “Technologically, I have no doubt that we’ll be able to achieve safe and reliable outcomes through pilotless air taxis. But I’m less certain about whether we’ll be able to convince passengers to board an aircraft which doesn’t have a person in charge with a vested interest in the outcome of that flight,” Dr Matt says. “It will be a major barrier, and it could take generations before that final jump can be made to total unmanned air taxis. Until then, a hybrid system is far more likely.”

He explains that as public confidence in the how the system will operate is fostered, humans will of course play a vital part in the air taxi system through a hybrid model that will continue to exist in some ways in the longer term. “The systems I describe are complex, and they will require monitoring by humans long-term, in the same way as other driverless transportation systems, like metro trains, operate today. Plus, the public are never going to tolerate an overnight transition into pilotless air taxi operations, so shorter-term, pilots will still be required. Though their role in that system will look very different.” 

The human element

In the meantime, AI technologies will also need to advance some, before these can be relied upon for autonomous drone operations. “For the next generation of autonomous systems, which process this information and work in collaboration with the human, the real challenge will to synchronise or harmonise that with the human,” Dr Ebbatson says. He alluded to past aviation accidents, such as the Boeing 737 MAX disasters, which boiled down to miscommunications between automated systems and humans, and an inability to easily override a system when necessary.

“The bigger picture is that automation has overall massively improved the safety outcomes of modern flying, because it gets rid of a lot of the weaknesses of human performance, which is consistency and speed of processing. But if we’re going to hand more autonomy over to the AI, we need to make sure if the human is going to still be ‘in the loop’ with the decisions that the computer is making, and how it’s coming to those decisions,” Dr Ebbatson says.

“I worked with a project in New Zealand which included an incredibly sophisticated air traffic control system that could automatically safeguard the separation of aeroplanes across the Pacific for 11 hours or more and gave aircraft a very specific trajectory which meant they would hit anything. But it didn’t show the controller how it’s making those decisions. All the controllers were able to do is basically say ‘yes’ or ‘no’ to the decision that the automation was offering. 

“And the problem with that was the controller ultimately remains legally and morally responsible for the separation of those airplanes, so you can’t just take away their visibility, and not inform them of how the automated system is going to do that. So we’ve really got to avoid doing that in our future systems. We want to keep humans monitoring the system, but we need to keep them in the loop properly.”

All this is to say that any future eVTOL system, including both vehicles and infrastructure, will rely on a complex hybrid system of advanced AI and human monitoring. So how exactly does this AI-supported, autonomous UAV air taxi network operate?

The nitty gritty

“As I said, central to all of this is going to be a backbone of data, which these vehicles will need in order to make sense of the world around them,” Dr Ebbatson says. These vehicles will need to not only collect data, but distribute it throughout the network, in order to keep the entire system afloat, he says.

For example, the vehicles will need high-quality onboard sensor systems, which allow them to not only navigate, but keep themselves separated from each other, buildings, and other aircraft in the airspace. The system will also rely on remote communications systems to relay information from the ground to each and every one of these vehicles, such as weather conditions, traffic conditions and changing passenger demand in certain parts of the network.

“Getting that information reliably to a vehicle is hugely important, and it involves making many, many tactical and strategic decisions very quickly, with certain decisions impacting the rest of the network and the rest of the airspace. So really, these decisions need to be made across the network, using data from all the vehicles,” Dr Ebbatson says. “This ecosystem will be incredibly complex, and again, you can start to see here how difficult it becomes to have humans attempt to manually monitor and control these operations.”

And yet, again, humans are still necessary to the operation of the system, just in different ways to previously. “The pilot’s role, for example, is going to become quite different,” Dr Ebbatson says. “Today, it’s a highly skilled profession that takes a significant amount of time and money to learn and perfect. But in the future, at the scale that an air taxi system could operate at, across various major cities at once, it would be basically impossible to continue to require that level of skill of these operations,” Dr Ebbatson says.

It marks another example of why automation will become so crucial to the air taxi system, he explains. These vehicles will be consistently operating under very clear, pre-determined routes and conditions, with all other necessary information fed directly to them through the network, Dr Ebbatson says. So, pilots will naturally “take more of a supervisory role” over the aircraft, as opposed to actively navigating and controlling the vehicle. 

A whole new world for pilots

However, the significant shift brought about by this practice is going to raise a lot of questions about the level of skill required to pilot these air taxis, particularly as they become more and more autonomous. First, you cannot realistically ask highly trained existing pilots, such as airline pilots, for example, to become supervisors onboard an eVTOL air taxi, and spend most of their days doing absolutely nothing at all. Second, an air taxi system would not be able to pay market rates for highly trained airline pilots. “If we’re going to suddenly end up with thousands of air taxis flying around, we’re going to need a lot of people to do this job, just like taxis. So the job in itself can’t demand high levels of skill, and training needs to be largely streamlined. Essentially, you need to make this job less skillful, because skill implies that you’ve got to train someone very hard, and it takes a lot of time and money to do that.”

Dr Ebbatson says that increased automation will allow for this transition to make piloting these vehicles a “less skilled” role than piloting a passenger airliner. “This is still going to be a skilled role, but with less training required in order to do it effectively. The job will get exponentially easier with the aid of automation, so it makes sense for it to then become a role that doesn’t require the extensive training that we’ve expected of pilots of traditional aircraft in the past,” he says.

The role of the human supervisor will come to play during “fringe” events and emergency scenarios that don’t let the vehicle track along its set corridor, or otherwise follow instructions from air traffic control. “Like most of aviation, 99 per cent of the time things operate routinely, smoothly and automation can easily handle that, and do it absolutely reliably. But it’s the edge cases that are complex.” So, instances where the vehicles need to divert suddenly, or land on an uncertain area, those will be the times that human intervention may be required in some form.

But ultimately, this too is a problem that complex AI systems will be able to conquer in due time, through carefully planned and pre-determined emergency contingencies, Dr Ebbatson says. “While I think we’re a little way off having vehicles which can sense their surroundings reliably enough to be able to autonomously operate into constrained sites in the urban environment particularly in a non-standard or emergency situation, I can see us getting there. The routine operations will be easy to reach reliable and safe outcomes. The abnormal situations are going to be where we take a lot of time to build confidence in the system. I think that the technology will be developed and will be able to answer to all of these issues, but funnily enough, it will come down to whether we, and the aviation regulators, can actually accept that.”