Dr Rafal Zbikowski on emulating flapping insect flight

The Defence College of Management and Technology at Cranfield University isn’t exactly your average university campus. The main entrance is guarded by people with guns, and rather than groups of drunken students soaking up the sun in between lectures every open space seems to be occupied by missiles, tanks, helicopters and other assorted pieces of military hardware.

With its austere buildings and neatly ordered courtyards, the campus, in Shrivenham near Swindon, looks every inch the fusty military establishment — and yet in between teaching and educating the next generation of soldiers, teams of researchers beaver away on a range of projects that appear to have more in common with film fiction than the current realities of warfare.

In perhaps the most esoteric of these initiatives, one group is engaged in the development of tiny flying robots that emulate the flapping flight of insects.

The project is headed by Dr Rafal Zbikowski, an engaging Polish control engineer whose team is developing a prototype micro air vehicle (MAV) which, it believes, could pave the way for a new generation of surveillance drones.

Zbikowski — whose work is funded by the EPSRC, the MoD, the US military, DARPA and NASA — told the Engineer that while UAVs and satellites are great for outdoor reconnaissance, there is a strong requirement for flying robots that can operate in confined and cluttered spaces within buildings, stairwells, shafts, tunnels and caves.

Such systems would, he said, be invaluable for indoor reconnaissance, urban warfare or counter terrorism missions. He added that the US military has even expressed an interest in using tiny flying vehicles to deliver small explosive charges for destroying individual computers, thus avoiding the need to bomb entire facilities.

There are also potential applications beyond the military arena. Tiny, manoeuvrable flying robots could, for instance, be used for pipe inspection in the chemical industry. This is currently carried out manually using mirrors mounted on sticks.

But the challenges of indoor flight are enormous. Vehicles must be capable of great agility at relatively low speeds. Their small dimensions mean that they must be extremely power efficient, and they must also be stealthy and quiet. Zbikowski said that silent flight will be particularly important as it’s likely that MAVs will be used in swarms — enabling them to cover large areas.

And while many existing reconnaissance devices are remotely operated, the prospect of controlling a swarm of robot flies around a cluttered environment without being able to see them simply isn’t feasible. The only solution, said Zbikowski, is to make them autonomous.

Before settling on insect flight as the model, Zbikowski considered a number of alternatives. He began by looking at fixed wing vehicles, but quickly concluded that they would lack the manoeuvrability required.

In terms of agility, tiny helicopters initially looked more promising, but this concept was dismissed on the grounds of their noisiness, low energy efficiency and the fact that when flying close to vertical walls, aerodynamic phenomena can cause helicopter flight to become dangerously unstable.

Insect flight, however, ticked all of the right boxes. Power efficient and honed to near perfection over some 300 million years of evolution, this method of flight endows insects with great agility at low speeds. Zbikowski gave the example of the lacewing fly which is capable of making a vertical upside down take off. ‘Nothing man-made can approach this — if we can reproduce this kind of performance our mission is accomplished.’

Zbikowski assumed that having identified the solution, the rest would be easy — but he was in for a shock. A visit to one of the UK’S leading experts in the field, Charles Ellington, professor of Animal Mechanics at Cambridge, revealed that very little was understood about how insects fly. So Zbikowski’s group found itself in the unusual position of working on an engineering project that is simultaneously helping push out the boundaries of biology.

‘Had I known what I was embarking upon I would never have done it,’ he said. ‘I was convinced that the fundamentals were largely sorted out on the biology and all we had to do was pick it up from there with appropriately qualified engineers.’

The team has so far developed a non-flying prototype that mimics the flapping motion of a hover fly. The design of the wing itself was the easy part, explained Zbikowski. Indeed, from a mechanical engineering point of view, insect wings are far easier to copy than birds’.

‘A bird wing has hundreds of bones, muscles and nerves and they can even control individual feathers and groups of feathers’ said Zbikowski. ‘Insects wings have no muscles. There are sensors but no actuators. All the actuation is done at the root and this simplifies things enormously as we don’t have to build a magic prosthetic device with hundreds of embedded motors.’

However, copying the unusual flapping motion of insect flight is somewhat more difficult. Insects fly by oscillating and rotating their wings through large angles while sweeping them forward and backwards and it has been tricky to replicate this motion mechanically.

The current solution involves a device — the spherical double scotch yoke — invented by Zbikowski’s group. This was developed by taking a single scotch yoke — a mechanism that converts up-and-down motion into rotary motion or vice versa — and crossing it with another scotch yoke to create a yoke that produces a figure-of-eight motion.

One of the keys behind the design has been a tool developed by Zbikowski’s Phd Students that is aiding the understanding of the aerodynamics of insect wings. ‘If you ask me to design a wing for a fixed wing aircraft there is a huge amount of knowledge, but here there isn’t,’ he said. His group has therefore spent a number of years generating a model that is able to predict lift and drag based on wing kinematics and geometry.

The biggest challenge of all is likely to be flight control. Very little is understood about how insects with tiny brains are able to perform flight control procedures and solve problems that would trouble a supercomputer.

Zbikowski explained that flies use only around 3,000 of their neurons for flight control, which equates to a rather paltry 3,000 transistors. ‘Your toaster has more computational power than this,’ he said. He proposes that they owe their aerobatic agility to a process described as sensor-rich feedback control. For while they may not have many neurons, flies are bristling with sensors.

He said that this contrasts with conventional aeronautics, which only measures a few factors — such as acceleration, altitude, and air pressure — and then employs powerful computers for flight control. ‘If you measure very little you have to compute a lot. I propose that insects do the opposite: instead of computing a lot they measure a lot and compute very little,’ he said.

Another big area of discussion is how to power such devices. One possibility is a miniature combustion engine, which is attractive because of both the high-energy density of fuel and the reciprocal motion generated by a piston engine.

However, a combustion engine would require a cooling system, as well as pipes, pumping equipment and a fuel tank, all adding complexity and weight to the system. The use of fuel would also mean that engineers would have to find a way of addressing the changing distribution of mass caused by shifting fuel levels. The most promising solution, and Zbikowski’s current preference, is for electric flight, and he hopes to be able to exploit the tremendous breakthroughs being made in electronic device design.

‘The whole world is working for me because there are so many portable gadgets. There is a gigantic applications pool and I don’t have to lift a finger. Every year batteries are improving.’

Clearly, with no flying machine to show for his troubles, and much fundamental research still to be carried out, the intelligent robot fly of countless science-fiction movies is some way off.

Zbikowski’s estimate is that, providing the funding doesn’t dry up, he could have something flapping around the laboratory in the next seven to 10 years. It may be a little longer, however, before swarms of insect-like micro air vehicles are deployed in the field.