Robotics researchers are increasingly turning to the animal kingdom for inspiration.
In a darkened room, a sinister buzzing sound begins, and flying mini-robots suddenly emerge from hiding. More and more of them appear, swarming across the room to reach their pre-assigned destinations. As the lights come up, they hover expectantly above their targets on tiny four-way rotors, ready to begin the mission.
...and that's when the robotic creatures start to play the theme to James Bond. Their precise control systems allow them to bob up and down playing the cymbal, keyboard, drums and maracas. One even carries out an autonomous fly-past of an electric guitar.
When Vijay Kumar of the University of Pennsylvania unveiled this video of his lab's flying quadrotors during a TED talk in February 2012, it gained a rapturous response. The robots' agility and apparent ability to cooperate was impressive indeed. But the tongue-in-cheek approach of the video, created by Kumar's team members Daniel Mellinger and Alex Kushleyev, was only too clear to the TED audience. The slick lighting and melodramatic choice of music played deftly on the way in which research into bug-size robots tends to be treated in the media - with coverage that is usually more sensational than real.
'Scary spy insects for the army' is how Michel Maharbiz, associate professor in the department of electrical engineering and computer science at the University of California Berkeley, characterises many media stories.
Even a quick search online reveals headlines that confirm Maharbiz's view'US Military AdmitsSpy Drones As Small As Bugs' 'Cyborg Army Moth Spy'... The sense of dismay in the scientific community is palpableby focusing on the fantastical, the media risks losing sight of an amazing story of success in learning to copy and harness nature's structures and skills for a huge range of tasks.
Kumar's quadrotors, for example, mimic the swarming behaviour of insects. His team's robots are only 20cm across, which increases their agility. But the really clever part is making them move purposefully in unison without hitting one another or crashing. To develop the skill of teamwork, Kumar looks to Aphaenogaster ants. These tiny creatures sense the behaviour of their neighbours, and without explicit communication, coordinate their work of carrying huge chunks of fig back to the nest.
The James Bond performance shows off this behaviour. Mellinger and Kushleyev had timed each quadrotor to be in a certain place at a certain time to make their rhythmic music - and the robots themselves worked out how to achieve this without hitting either an obstacle or a fellow robot. While flying in formation, the robots can also work out how to squeeze through restricted spaces without any centralised coordinator, as well as working together to build simple structures. You could imagine these ant-inspired skills might see quadrotor teams one day checking an earthquake-damaged building, or even working to lift rubble off injured people.
And it's not just ants. Scientists are also looking to bees, cockroaches and dragonflies for their hive intelligence, their resilience and their power of flapping flight. What's even more intriguing is that a parallel strand of research is showing how insects themselves can be harnessed to function under remote control.
Biomimetics - the copying of nature's solutions to technical conundrums - has incredible potential. If animals can supply an optimum answer to a problem such as locomotion or sensing, there's no particular benefit in trying to reinvent the feeler, the foot or the flipper for your robotic creature.
Indeed, fins and flippers were one of biomimetics' early success stories. Robotuna, developed at MIT, was a prototype swimming robot that replicated the way a real fish moves through the water - with a view to making underwater vehicles more efficient. The articulated robotic tuna did indeed use less energy to swim than other robotic submarines, as well as being more manoeuvrable.
Robolobster was built at Northeastern University, and was designed to be able to negotiate the terrain at the bottom of the sea by flexing its shape-memory-alloy muscles and respond autonomously to its environment. FLAPS, developed at Heriot-Watt University, was a prototype rippling, pneumatic fin that mimicked the propulsion system of a seahorse.
Penguins have inspired more recent developments for efficient autonomous swimming. German company Festo has employed new studies on fish fins which show that, unexpectedly, they respond to a lateral force by bulging out instead of bending away. By recreating this high-efficiency effect in three dimensions, Festo has created a torso that can move in any direction, allowing the Aquapenguin to manoeuvre in cramped areas and even swim backwards, just like the real creatures.
On land, biomimetic robots have enjoyed a field-day too. Boston Dynamics, a spin-off engineering company from MIT, unveiled BigDog, a powerful-looking creature with legs articulated like those of an animal. BigDog can travel 12 miles without refuelling, walk across rubble, snow and in water, and carry up to 150kg.
At Stanford University, attention has focused on mimicking geckos' ability to scale walls. Stickybot III is the latest iteration of their robotic creatures. Its feet pads are made of rubber, which is cast in a mould to produce an array of microscopic wedges that adhere to a surface via the Van der Waals force. Unlike sticking and unsticking a piece of sellotape, researcher Mark Cutkosky describes the sensation as "being able to hook and unhook yourself from the surface".
Cutkosky and his team were creators of an early robotic cockroach, developed with the idea that insect-like creatures could overcome the problems encountered by wheeled robots - for example on missions to Mars. Sprawl, developed with Robert Full at Berkeley, had six legs propelled by tiny compressed-air pistons.
Search and rescue
Full's group made the recent discovery that cockroaches, lizards and geckos all share an ability to rapidly invert themselves out of sight - by swinging acrobatically under any handy ledge. Colleagues in the Biomimetic Millisystems Lab at Berkeley experimented with their robotic roach called DASH - the Dynamic Autonomous Sprawled Hexapod - and found that with a little tweaking, it could do the same. A possible application of this trick might be a highly mobile search-and-rescue bot particularly adept at negotiating urban disaster zones.
A different bug is on the crawl at Harvard's Wyss Institute for Biologically Inspired Engineering. Robert Wood and colleagues have created a 12-legged, segmented robot based on a centipede. Unlike the rigid bodies of roach-based robots, this creature has flexible connections between its segments, and ripples along. The researchers hope to study the way flexibility and body undulations help the creature to move.
If you look overhead, you find one of the areas of greatest development in robotic biomimicry. At Delft University, researchers have been working on Micro Air Vehicles (MAVs) inspired by dragonflies. Their DelFly prototype has now gone through three iterations, shrinking from 50cm to 28cm to only 10cm (DelFly Micro, weighing only 3g). At each size the team has shown the bug can still carry an onboard camera and fly, untethered, in an indoor environment. Researchers would like to reduce the scale still further"In the future we hope to develop a MAV that is the size of a fruit fly," says Bart Remes from the team, although he admits this is decades away.
And back at Harvard, the Mobee project is taking flight. Bees are unparalleled in their ability to fly and hover while carrying a heavy load. Researchers in Robert Wood's Microrobotics Lab, who saw their first successful flight of a life-sized robotic fly in 2007, have now been learning from bees. Compact power-sources are one of their research interests. Another is a sensing system, mimicking the bee's eyes and antennae. Finally, they are learning from the bee's skills at coordinating activity with its fellows in order to bring home the nectar.
One breakthrough the team has made is in replication. By learning how to manufacture the Mobees using a sandwich of materials that are cut out using a laser and then folded into shape, origami-like, they could conceivably create the swarm of creatures beloved of headline writers - but which actually brings search-and-rescue or monitoring applications within budget.
Scientists are making huge strides in this increasingly tiny science - but some issues remain with robotic creatures, particularly when they take to the skies. To power a flying robotic creature is no mean feat for more than a few minutes, and so some researchers have approached the problem by turning it completely on its head.
In the 1990s, researchers began to wonder whether it might be possible to learn to control real-life insects, thus harnessing directly their powers of locomotion or flight. Might it be possible to control the signals received by an insect's muscles or nervous system, and thus create a remote-controlled biological robot?
At Cornell University, Amit Lal and Alper Bozkurt worked on the idea at a very fundamental level. They demonstrated that microelectronics-based systems could be implanted into moths during their metamorphosis, integrating the technology directly with the creature. This was the first step towards the goal of creating 'cyborg insects' - perhaps a rather sensational name, but one that seems to have stuck.
Under a programme organised by Lal for the US Defense Advanced Research Projects Agency (DARPA), researchers at several universities investigated hybrid insects.
Michel Maharbiz, by then at University of California, Berkeley, worked with flying beetles, using their natural response to light and dark to encourage them to fly or stop flying. By stimulating a beetle using an electrode implanted into the optic lobe, the team could make it flap its wings and assume a flying posture. To direct the beetle, a pulse to the flight muscles on left or right would make it turn.
Bozkurt continued his work with moths, but is now exploring terrestrial insects with his own team at North Carolina State University. He can control a cockroach's direction by stimulating the creature's antennae, as if it had brushed an obstacle. After working out a pulse strength that didn't damage the roach, he found that the greater the electrical pulse delivered from a tiny backpack, the more sharply the creature turned. "We recently demonstrated cockroach navigation with an aim of rescuing victims after natural disasters in the future," Bozkurt explains. The work is now funded by the National Science Foundation.
And a project with moths at MIT also reported success earlier this year, with researchers showing they could control the creature's turning to the left and right using a wireless signal. The moth had a flexible neural probe attached to its central nervous system, with five electrodes to stimulate different bundles of nerves. Although the customary headlines about cybersnooping prevailed, the team actually saw the research as promising for neurobiologists seeking to stimulate human nerve bundles and rehabilitate people who had lost mobility after a stroke.
Perhaps biomimetic creatures and adapted insects will one day save lives and offer solutions to scientists wanting to monitor situations on land, at sea and in the air. The possibilities seem endless, and worth looking beyond the hype to find.