Imagine an insect that can alight along a pipeline to sniff out gas leaks, swoop beneath plants to detect pests and disease, and slip into tiny spaces to find disaster survivors. Now imagine holding a laser that can control its every move.
With the rapid advances in drone technology spanning the 20th century, it should come as no surprise that miniature flying robots are on the horizon: Between now and 2020, Goldman Sachs' forecasts a $100 billion market opportunity for drones, helped by growing demand from the commercial and civil government sectors.
What is surprising is that it has taken researchers more than two decades to finally come up with a fully autonomous version. That's because the electronics needed to power and control the wings were so heavy that, until now, flying robotic insects had to be tethered to a wire attached to an external power source.
Yet a team of engineers at the University of Washington, led by assistant professor Sawyer Fuller, were able to figure it out. Relying on funding from UW, they created RoboFly, a robo-insect powered by an invisible laser beam that is pointed at a photovoltaic cell, which is attached above the robot and converts the laser light into enough electricity to operate its wings.
Because the laser alone can't provide enough voltage to move the wings, the team designed a circuit that boosts the 7 volts coming out of the photovoltaic cell up to the 240 volts needed for flight. To give RoboFly control over its own wings, the engineers added a microcontroller to the same circuit, which acts like its brain, according to Vikram Iyer, a doctoral student in the UW Department of Electrical Engineering who is part of Fuller's team. "It tells the wings things like, 'Flap hard now' or 'Don't flap,'" he said.
Fuller received both his bachelors and masters degrees in mechanical engineering from MIT and later his Ph.D. in bioengineering from Caltech. He now directs the Autonomous Insect Robotics Laboratory at UW, which works to advance insect-scale robotics engineering and better understand the capabilities of insects. "Robotics on the insect scale have a lot of constraints. That's what interests me," he said.
By constraints he means "fly-sizing" things like scanning lasers, cameras and range finders. "A lot of the sensors that have been used on larger robots successfully just aren't available at fly size. Radar, scanning lasers, range finders — these things that make the perfect maps of the world, that things like self-driving cars use. So we're going to have to use basically the same sensor suite as a fly uses, a little camera," Fuller said.
Fuller became inspired by insect robotics 20 years ago when he started observing a group at UC Berkeley who were trying to create a "micromechanical flying insect" that weighed less than a paper clip and could lift off the ground and hover.
"They started having success, but these little flying robots needed a wire going up to them." Later he got involved with a series of other advances. "We got them to fly under control rather than just take off; we put sensors on board and made them land and do other things."
Now 41, Fuller and his team is one step closer to creating a fully autonomous robo-insect: For now, RoboFly can only take off and land. Once its photovoltaic cell is out of the direct line of sight of the laser, the robot runs out of power and lands. But the team hopes to soon be able to steer the laser so that RoboFly can hover and fly around. They are currently working on more advanced brains and sensor systems to help the robots navigate and complete tasks on their own, Fuller said.
"For full autonomous I would say we are about five years off probably," he said.
According to Goldman Sachs, the three largest industries for drones are construction, agriculture and insurance, with the total current market value for each at $11 million, $5 million and $1.4 million, respectively. Fuller sees many opportunities for robo-insects that the larger drones can't handle. The application he sees as most promising for flying robotic insects is micro-agriculture. "Robotic flies will have a part to play in terms of their ability to fly around and really monitor the environment — things like humidity, state of disease — in a very detailed level, flying down, in and above the plants to look for disease or pests and to monitor air quality."
More from The Edge:
Fuller also plans to add odor sensors so RoboFly can sniff out natural gas leaks or fuel leaks. "One of the things that flies do really well is look for smelly things. Our robotic flies can do that pretty well. And if you can find leaks, then you can patch them up more readily and reduce greenhouse emissions," he said.
In addition, Fuller said RoboFly's mere size makes it a winner over the larger drones for two very big reasons: cost and battery life. "Larger drones have a pretty limited battery life. Maybe 30 minutes at best. You are going to need to spend a little bit longer in the air to find a leak. You also need a much more powerful laser to power a large drone. We can deploy a lot of them all at once for the same cost as one large drone because of the low materials cost. So you could have a hundred of them doing the job of one or two big drones."
"There's no reason I see why it would inherently cost any more than the little radio-controlled drones that you could buy that have become crazy cheap now — helicopters that are a couple inches across that you can now buy for $20."
But the return Fuller sees ahead may be enormous: He and the team have been reaching out to ExxonMobil to see if they have any interest but as yet have not heard back. "We would definitely be interested in a collaboration like that, for sure," he said.
While there are many opportunities for robo-insects, there is always the issue of security. "I think it's really important to publish everything we do openly so that its public domain, so people can see what's going on and know where this technology is going. Part of my job is trying to make sure this gets used for good things," said Fuller.