Bipedal robots have been a tough ask for engineers. Creating a bot that’s steady, self-balancing, and able to adapt to uneven terrain (one of the main advantages of going bipedal in the first place!) is a tough ask. But, as this newly unveiled bot from Agility Robotics proves, we’re getting good at it.
The bot’s name is Cassie, and, as reported by IEEE Spectrum, it comes from a fine lineage of bipedal robots. Agility Robotics is a spinoff company from Oregon State University, and the firm’s researchers previously created the ATRIAS robot. (You may remember ATRIAS from a video of it playing a slightly one-sided game of dodgeball.)
“ATRIAS was the first machine to demonstrate human-like gait dynamics and implement spring-mass walking, but it was not a practical machine for any use other than science demonstration.” Agility Robotics co-founder Jonathan Hurst told Spectrum. (“Spring-mass walking” basically uses the elasticity of springs to create a passive mechanism mimicking human muscles.)
As well as improving this mechanism, Cassie also adds a 3-degrees-of-freedom hip joint that allows it to be steered more easily, and powered ankles that mean it doesn’t have to jig from foot to foot to stand still. (It can just… stand.) A possible final design for a consumer model of Cassie is mocked up below:
But what could such a bipedal bot be used for? Well, walking around on two feet may be complex in engineering terms compared to, say, a wheeled robot, but it means the bot can go pretty much anywhere humans can. Rocky ground? No problem! Stairs? Taking ‘em one by one. Agility Robotics suggests their tech could be used to make search-and-rescue bots; to help improve prosthetic limbs or exoskeletons; and, presumably if the bots become cheap enough, delivering packages. They just need to remember to give it a telescoping stick so it can poke doorbells.
Oh, and because the video above shows a self-balancing robot, here’s the traditional kick-to-the-abdomen all such bots seem to be subjected, too:
Just wait till it learns to kick back.