This could be the future of robotics — and it kind of looks like a chicken.
Agility Robotics unveiled a bipedal robot called Cassie this week. The company, which spun out of the ATRIAS project at Oregon State University, is focused on “legged locomotion” and hopes to someday engineer robots that can walk just like people.
That should be incredibly useful in a wide range of applications — but it needs some more work before then.
Cassie was built using a 16-month, $1-million grant from the U.S. Department of Defense’s Advanced Research Projects Agency (DARPA). The custom design took cues from animal morphology and behavior, including a hip joint with three degrees of freedom and flexible ankle joints, just like a human’s leg. This makes it much more stable than the team’s previous designs, whose motor-heavy systems were inefficient.
Cassie can stand, squat and balance on its own and “take a pretty good fall without breaking,” which was essential during the testing phase. According to Cassie’s makers, most other bipedal robot designs struggle to walk a straight line and fall over at the slightest disturbance.
The bots could someday be used to assist soldiers in the field, for emergency services like search and rescue missions, or even for tasks like grocery deliveries — just like the Vespa cargo bot that also debuted this week and looks a bit less creepy.
Unlike the Vespa, though, Cassie’s two-legged design will give it access to places wheeled bots can’t go. Agility Robotics’ chief technology officer Jonathan Hurst imagines a future filled with walking robot assistants that could someday free people from monotonous errands like grocery shopping.
For now, the bot is still a work in progress, although a commercial model is launching this month. Hurst told IEEEE Spectrum additions like arms and advanced sensor systems are coming soon to make the robot more autonomous, and the team is also looking to implement a VR-style telepresence for remote control.
With better user control systems and AI, Cassie can move closer to its full potential — but that’s still a ways down the road. For now, we’ll call it a win when the bot can master stairs on its own.