DASH to the Next Gen of Robots: Small, Cheap, and Feral

by Gordy Slack

What do you get when you cross a gecko’s climbing ability with a cockroach’s flexible durability? In the Biomimetic Millisystems Lab at UC Berkeley, you get a lightweight, resilient, and sensitive robot that can go almost anywhere. And if one or two get squashed, it’s no big deal: they are super cheap!  

Using compliant fiber board as structural material, and a single main driver motor, the DASH robot is capable of 15 body lengths per second on flat surfaces. The structure is resilient and survives ground impact at terminal velocity of 10 meters per second.

With their gecko-inspired, adherent feet, the Little Robot That Could, called DASH (which stands for Dynamic Autonomous Sprawled Hexapod), could deploy sensors for many other purposes, says UC Berkeley Professor Ron Fearing, Director of the Biomimetic Millisystems Lab.  For example, they might have helped anticipate the failure last October of a repair to the San Francisco Bay Bridge, in which a heavy steel crossbeam came loose and hit three vehicles.  "It would have taken about $10 worth of sensors to show that the bridge was vibrating abnormally," Fearing says.

"Eventually," he says, "we will be able to dump a bucket of these robots on the bridge and program them to go and adhere to various spots that would be tricky for humans to reach. It would be safe, the labor cost is low, and you would not have to close the bridge. If a sensor falls, no one will get hurt since they weigh only a few grams."

Paul Birkmeyer, the graduate student who is spearheading DASH’s design, was first inspired to develop inexpensive and durable robots that could be dumped by the hundreds onto disaster sites. These could work, even before the adhering gecko feet are perfected, by releasing them at the highest point on the damaged building, and then letting them work their way downhill. The robots could assist rescue workers seeking out survivors of natural disasters. A batch of them, equipped with CO2 detectors, could thoroughly cover a dangerous building and investigate spaces too narrow for human rescuers to enter.

Birkmeyer, a graduate student at UC Berkeley in Electrical Engineering and Computer Sciences, looked to nature for mobile inspiration. The six-leg design exhibits the "spring loaded inverted pendulum" used by cockroaches to propel themselves forward and climb over obstacles, says Fearing. "That insight was Birkmeyer’s key to getting DASH to move so nicely," he says.

"The age of robots will have arrived," says Fearing, "when you can buy one for ten dollars that does something really useful." And the next-generation DASH, the ten-dollar cardboard gecko-roaches, may mark that day.  The cheap little bots could also do more mundane domestic jobs like washing hard-to-reach windows, sweeping cobwebs out of high corners, or conducting search-and-poison missions on, say, rodent nests.

The electronics in DASH takes advantage of the tremendous progress in electronics and MEMS sensors of the last 20 years. Each robot carries a stamp-sized battery and a single motor and can easily be equipped with other electronics like a cell-phone camera, standard sensors, and Bluetooth wireless. But the big innovation here is in the simple way the robot moves and, eventually, the way it will climb.

"Two of the three main ingredients for a good, inexpensive, sensing robot are already in place," Fearing says. "We have powerful, light, inexpensive computation. And we have great affordable sensors. All we need now is to solve the mobility challenge; we need something inexpensive that moves reliably over all kinds of terrain."

The next step is to give DASH the power of verticality. And for that, Fearing’s team is looking for inspiration at the fancy footwork of geckos.

The basis for geckos' adhesive properties is in the millions of micron-scale setae on each toe of the gecko form a self-cleaning dry adhesive. The tip of each seta consists of 100 to 1000 spatulae only 100 nanometers in diameter.

Each gecko toe employs a "complex hierarchical structure to achieve adhesion," says Birkmeyer. A single toe has dozens of flap-like structures, known as setae; and each seta has hundreds of thousands of stalk-like structures. Each of those in turn has roughly 1,000 very thin hairs.

So far, Fearing’s lab's has only replicated the lowest level of the hierarchy, the tiny hairs. That gives great adhesion on smooth surfaces under very controlled conditions. But synthetic versions of the larger structures, the flaps and the setae, which hold onto rougher surfaces, are still under development. In addition to synthesizing those other layers of gecko tenacity, Birkmeyer is also developing a foot-ankle combination that would automatically conform DASH's foot to a surface to keep even the finer gecko material in contact.

"Intuitively, running up a wall feels very different than running over a flat surface," says Birkmeyer. "And studies on cockroaches and geckos show they use very different foot motions for climbing  and running. The new foot would allow the adhesive hairs to sort of fall into place as the foot hits the wall," he says.

After they get the gecko feet to adhere, allowing the robots to scale walls, the team will then work on DASH’s ability to walk on uneven materials such as gravel, sand, and grass. Each situation presents its own set of challenges, says Birkmeyer, but in each case, nature has insights to offer.

The inspiration goes both ways, says Fearing, who collaborates with biologists like Robert Full, UC Berkeley Professor of Integrative Biology. Full did the basic animal research that went into DASH's gait and was also a principal player in the gecko adhesion research.

As in nature, Fearing's team also uses trial-and-error to advance. Because their fabrication methods use a precision, computer-guided laser that cuts DASH from a single piece of paperboard, like a complex paper doll, it is relatively easy and inexpensive to make changes and try new things. But the mutations that take place in this lab are hardly random.

"Sometimes you learn things from the synthetic system that cause you to ask questions about the natural system; a robot may have a weird behavior that we did not see in the natural system. Then, when we go back and look more closely, we find that evolution had to find a solution to the same problem. But the natural design may work so smoothly that biologists might not have appreciated what may have been a key design feature," says Fearing.