Researchers are using robotic simulators to explore the possibilities of autonomous aerial refuelling.
Earlier this year, two unmanned aerial vehicles (UAVs) – Northrop Grumman’s Proteus and NASA’s Global Hawk – flew as close as 40ft (12.2m) apart at an altitude of 45,000ft (13.7km). The test flight was intended to be a first stepping stone on the path to entirely autonomous aerial refuelling. But, in truth, it was very much a tentative baby step. 40ft might not seem like a long way, but it’s when an aircraft crosses this threshold that the pilot’s skills are tested to the maximum.
A recently launched research project involving researchers at Bristol University and UK aerospace company Cobham is attempting to close that gap with the help of some novel robotic simulations.
The project is actually one facet of the wider ASTRAEA (Autonomous Systems Technology Related Airborne Evaluation and Assessment) programme, which aims to open up non-segregated airspace for UAVs and, thus, a wider range of applications, many of them non-military.
One of the key quoted advantages of UAVs is that they can remain in the air for longer periods, unburdened by the physiological limitations of human pilots, who tire and by law must be relieved by a fresh crew after a period. However, unlike their piloted counterparts, UAVs cannot, at the moment, refuel mid-air and so will inherently be limited by the amount of fuel they can carry.
’If unmanned aircraft systems [UAS] are going to fulfil their complete potential, there’s a good chance they’ll have to refuel,’ said Richard Bourne, a former naval officer and now programme manager of research and technology at Cobham Mission Equipment.
’That may seem bizarre in the non-military sense but it’s where refuelling started – Cobham’s initial concept was to support Imperial Airways, enabling its flying boats to get across the world without having to land and refuel regularly.’
Indeed, conventional air-to-air refuelling was originally explored in a civilian setting, with the US interested in achieving trans-Atlantic flights for a faster postal service to Europe, while Britain wanted to service the Empire with its flying boats.
However, the cost of the equipment and the complexity and skill involved in actually executing the procedure meant that aerial refuelling to this day has been confined to military operations. So, ironically, part of the ASTRAEA programme is to bring the technology full-circle.
’Obviously the challenges of removing the human from the process of refuelling is considerable. It’s only when you start picking the problem apart that you realise just how smart humans are and how they are able to absorb information, translate that into action and react accordingly,’ Bourne said.
For cost and practicality purposes, the project will work on the assumption that UAVs will use existing refuelling tankers employing the ’drogue-and-probe’ method. Most of the development will therefore go into the autonomous receiver aircraft, which will make positional decisions.
’You’ve got to emulate the eyes, arms and brain of the pilot to sense where the tanker is, where the drogue is, how they relate to each other, where you are, then how to move in. Once you’ve connected with the drogue it becomes part of the receiver, so you have to maintain formation with the tanker,’ Bourne said.
So the first aspect of the project is to understand what actually happens during a successful air-to-air refuelling manoeuvre – the various physical forces at play and the decisions that the pilot makes.
As one of the early pioneers of aerial – refuelling technology, Cobham enlisted the help of Bristol and Cranfield universities to assist it with wake, atmospheric and aerodynamic modelling, and some of the UAVs’ existing control systems.
Having established a ’synthetic environment’ incorporating all of these aspects, the team needed an arena to test out possible hardware and software solutions.
As every engineer knows, computer simulation is no substitute for the real thing, but clearly it wouldn’t be practical to test out new solutions in real UAVs.
Cobham therefore commissioned a unique robotics system to be installed at Bristol University as a surrogate for the real thing. It comprises an exact copy of the probe that any recipient UAV would have, mounted on moveable head, which in turn is placed on a 10m-long rail rack – replicating that vital distance where the aircraft come together. The other robot is an exact replica of the shuttlecock-shaped drogue found on most tankers, which is also able to move as if being trailed on a hose.
Researchers at Cranfield University are also developing bespoke sensors to aid the recipient UAV in locating its spatial position in real time. Versions will be fitted to the robotics platform at Bristol, then tested and honed as part of the overall hardware product.
’The first task is to get the end of those robots to mirror as closely as possible what the synthetic environment is doing – that interface to the real world, which brings very difficult technical problems,’ said Bristol’s Dr Tom Richardson, who is leading the research effort.
’You need to get those right, so the answers you get out of the flight-simulation control and hardware and loop testing are going to be right.’
There’s something strangely hypnotic about watching the probe oscillating in the simulated ’wind’ while slowly closing in on, and matching the movement of, the drogue, like some sort of elaborate courtship dance.
Having received the system back in February, the team is still verifying the synthetic environment, running numerous simulation scenarios that UAVs might encounter up at real altitude.
“We can’t design a fantastic solution but we can say that we want full control of the aircraft”
Clearly the team has the best possible technology available to tackle the problem and it’s tempting to speculate that it might simply be a case of number crunching to develop the correct algorithms and tweaking various bits of hardware.
But there are still many problems that may not have definite solution. Cobham found that pilots can often ’anticipate’ the behaviour of the drogue and adapt based on experience. This is something very difficult to imbue a machine with.
For example, occasionally when the recipient aircraft approaches a tanker, it creates a bow wave with enough force to shift the drogue. Richardson will be supervising a PhD project specifically to see if bow-wave effects can be predicted, although it’s certainly not a given.
Eventually Richardson hopes to have some sort of product, most likely a combination of hardware and software, which he will hand to Cobham to be fitted on a candidate UAV aircraft.
’We are doing this constantly, bearing in mind that we have to work with existing aircraft. It’s highly likely that whatever aircraft this will go on will have existing hardware and software that we need to interact with. We can’t design a fantastic solution, but we can say: “Okay, we want full control of the aircraft”.’ Richardson said.
back story auto pilot
The ASTRAEA project could lead to UAVs performing search and rescue missions
The £62m ASTRAEA programme gathers seven companies – BAE Systems, Cassidian, Cobham, QinetiQ, Rolls-Royce, Thales and AOS – in an effort to open up regular airspace to UAVs.
’You can’t fly UAVs in non-segregated airspace, they have to be in their own cleared piece of airspace with everybody else excluded because of the difficulties of preventing them from bashing into other people, for example,’ said Richard Bourne of Cobham.
But if UAVs are to fly and operate in regular airspace, they will have to comply with normal regulations. To do that will require the development of technologies that effectively remove the pilot from the cockpit to put him on the ground.
’The classic one is sense and avoid – the pilot’s responsibility to look out of the window and make sure he doesn’t drive into anything. You’ve got to emulate that capability,’ Bourne said.
Ultimately it could lead to UAVs performing search and rescue, remote sensing, cargo carrying and mapping.