In the aftermath of a major disaster, the last thing you want to do is send first responders running into mounds of unstable, potentially dangerous rubble. But right now, if rescuers want to look for survivors, measure radiation levels, or just see what’s going on inside, they don’t have many other options. Nuno Martins, an associate professor of computer and electrical engineering, and his fellow researchers at the University of Maryland Robotics Center think the best way to solve this problem is with robots. Hundreds of tiny, ant-sized, autonomous, communicating robots.
Their project, funded by the National Science Foundation, aims to create a small army of extremely simple electronic search-bots that can climb in between tiny cracks, spread themselves out in a given area, and relay information back to a team of humans waiting safely outside. The bots will look for things like radiation levels, airborne toxins, life signs from survivors, and generally get the lay of the land. Then, depending on the distance between the reconnaissance bots and home base, they can use adapted wireless technology to send signals through relay bots that have placed themselves strategically to capture and resend information.
It’s not an easy task. The bots’ limbs and wireless communications equipment have to be engineered to work on very little power. Martins is focusing on eliminating complexity as much as possible. “The algorithm has to be simple,” he says. “You cannot have a very powerful computer on a very small footprint with limited battery power. You have to be smart with design.”
To achieve his goals, Martins is taking a hybrid approach to building the bots’ brains. In some cases they may rely on standard digital processing—using a microprocessor and 1s and 0s like a computer—and in others the robots will have to use tiny mechanical and electronic switches to perform computations. He’s hoping this will make the tiny power-light army better at thinking and troubleshooting when they encounter a situation they haven’t experienced before.
He’s applying the hybrid approach to his team as well. One member specializes in low-power analog circuitry. Two others focus on micro-motors that allow the robots to move. Others are working on making the tiny legs energy-efficient. “There’s a lot of theoretical work that needs to be done to understand what kind of performance we should expect. You need to have an idea if [an error] is due to the limitation of the algorithm or if it’s due the limitation of the hardware,” he says. For example, if a robot takes longer then expected to perform a certain task, like walking or turning, there could be two explanations. One is that its programming doesn’t work correctly. Or, if the algorithm predicts the task should be happening more quickly, then it’s possible the problem is with the movement mechanisms.
Right now the bots are in the prototype stage. Martins’ team has managed to get them down to about 1.5 square centimeters at a cost of $20 to $30 each. Once completed, he hopes the robot army can expand beyond rescue operations to other jobs like inspecting architectural frameworks and even performing surveillance for the military.
The next logical step in advancing the concept is figuring out how to make the antbot army fly. Martins has recruited University of Maryland aerospace engineer Sean Humbert to assist with the task—he already has a few prototypes that attempt to deal with the problem of wind gusts. “You increase the difficulty significantly when they fly,” says Martins.
Ultimately, Martins and his team aim to have the ability to take the tiny brains and actuators they’re building and use them to turn any group of tiny robots into an autonomous system. Sounds great—except, as we all know, that’s how Skynet starts.