eight years ago, The Pentagon’s Defense Advanced Research Projects Agency staged a agonizing contest of observation that involved robots slowly (and often failing) struggling to perform a series of human tasks, including opening doors, operating power tools and driving golf carts. Clips of them stumbling and stumbling during the DARPA Robotics Challenge quickly went viral.
Today, the descendants of those unfortunate robots are much more capable and agile. Many startups are developing human beings who claim that, within a few years, they can find work in warehouses and factories.
Jerry Pratt, senior research fellow at the Institute for Human and Machine Cognition, a nonprofit research institute in Florida, led a team that came second in the DARPA Challenge in 2015. He’s now a co-founder of Figure AI, a company that builds a robotic robot designed for warehouse work announced today An investment fund of $70 million.
If the DARPA challenge were run today, Pratt says, the robots would be able to complete the challenges in about a quarter of the 50 minutes it took the robot to complete the course, with few accidents. “From a technical standpoint, a lot of enabling technologies have come out recently,” he says.
More advanced computer vision, made possible by advances in machine learning over the past decade, has made it much easier for machines to navigate complex environments and perform tasks such as climbing stairs and grasping objects. More energy-dense batteries, produced as a result of electric vehicle development, have also made it possible to pack enough juice into a humanoid robot to move its legs fast enough to balance dynamically—that is, to stabilize itself when slipping or miscalculating a step, as humans can.
Pratt says his company’s robot is taking its first steps around a fake warehouse in Sunnyvale, California. Figure CEO Brett Adcock believes it should be possible to build a humanoid for the same cost as making a car, provided there is enough demand to ramp up production.
If Adcock is right about this, the robotics field is nearing a defining moment. You’re probably familiar with the dancing Atlas humanoid robots that have been gaining a cult following on YouTube for several years now. Manufactured by Boston Dynamics, a pioneer in bipedal locomotion that made some of the humanoids used in the Darpa competition, it showed that making robots capable of human form is possible. But these robots were expensive—the original Atlas cost several million dollars—and lacked the software to make them autonomous and useful.
Shape isn’t the only company betting that humanoid robots are maturing. Other applications include 1X, Apptronik, and Tesla. Elon Musk, CEO of Tesla, paid a visit to the original Darpa Robotics Challenge in 2015. The fact that he’s now eager to build a humanoid himself suggests that some of the technology needed to make such a machine is finally viable.
Jonathan Hirst, a professor at Oregon State University and co-founder of Agility Robotics, was also up for a DARPA challenge to give a demonstration of an animatronic robot he’d built. Agility has been working on two-legged robots for a while, but Hearst says the company has taken a physics-first approach to locomotion rather than copying the mechanics of human limbs. Although its robots look like humans, their legs seem to be inspired by those of an ostrich.