The Pentagon’s pet research agency wants robots that can reconfigure themselves on the fly as mission requirements change.
By Michael Peck
The U.S. military’s future robots may reshape themselves like Transformers – those popular shape-shifting robots from Hollywood movies and toys.
But DARPA (Defense Advanced Research Projects Agency) doesn’t want Optimus Prime as entertainment. The Pentagon’s pet research agency wants robots that can reconfigure themselves on the fly as mission requirements change.
“DARPA envisions future military operations where small robotic systems – enabled by on-board autonomy (navigation and perception) – may need to self-reconfigure (autonomously change shapes, modes, form, or function) to operate independently in unknown, highly complex environments,” according to the agency’s project announcement, titled “Self-Reconfigurable Modular Ground Robots.”
This means robots that can function in environments so difficult that humans fear to tread there. And not just function, but perform what DARPA terms “complex maneuvers.”
“An example of a complex maneuver could be: move over rocks/loose dirt, traverse a ditch, move through trees/bushes, mount a curb, and deliver a package in a precise location and orientation,” DARPA explains. “Designs should consider system approaches that are compact, lightweight, low power, low cost, and appropriate for use as expendable assets, when necessary.”
While Optimus Prime is depicted as a 20-foot-tall bipedal robot that can instantly turn into various machines such as a cannon-armed fire engine, DARPA’s goals are more modest. The agency foresees machines that are less than 2 feet tall, weigh less than 25 pounds, can travel at least one kilometer and carry a payload of 2 to 10 pounds.
This sounds less like a giant combat robot and more like something that Amazon.com would use to drop off your package. Which is exactly where DARPA seems to be going with this.
The goal doesn’t seem to be developing a clever robot so much as a versatile one that’s smart enough to autonomously reshape itself into the optimum configuration for traversing uncharted terrain. “For example, as a robot transitions from rocky terrain to sand, it may sense the difference in terrain to prompt a reconfiguration,” DARPA notes. “Or, a robot may need to change configuration to shift from a crawling mode to a climbing mode.”
Thus, the robot will have to capable of recognizing the environment that it will need to traverse, and then know how to change itself into the optimum configuration. And some of those environments will be challenging. DARPA wants a design “capable of operating in a diverse set of terrains and situations. These may include rivers, lakes, subsurface tunnels, rocky, forested, highly sloped, and uneven terrains.”