Cassie, the bipedal robot now has a world record for 100m dash

Cassie, the bipedal robot built by the researchers at Oregon State University (OSU), has set a Guinness World Record by clocking 24.73 seconds for a 100m dash. The robot started from a standing position, sprinted, and then returned to its starting position without falling, a university press release said.

Quadruped robots might all be the rage on the military side of technology, but the bipedal gait has all the attention on the academic side. Working on bipedal robots requires mechanical engineering, robotics, and computer science expertise. It has taught researchers a lot about locomotion on two legs, something we take for granted.

"Starting and stopping in a standing position is more difficult than the running part, similar to how taking off and landing are harder than actually flying a plane,” said Alan Fern, a professor of artificial intelligence (AI) at OSU, who has been involved in developing Cassie since 2017.

How was Cassie trained to run fast?

The researchers crammed the equivalent of a year of simulation environment into a week to train Cassie to run fast. The process is called "parallelization" where multiple processes and calculations happen simultaneously, which allowed Cassie to go through multiple training experiences simultaneously.

Cassie has been taught a wide spectrum of gaits, but as researchers were training it to pick speed, the bipedal robot quickly optimized a pace strikingly similar to how humans run.

The more significant challenge for the researchers was to ensure that Cassie completed the feat from a free-standing position and ended it in the same place without falling. It does not help, though, that Cassie isn't equipped with any external sensors or cameras, so the bipedal robot is always running blind for all intents and purposes.

A trailblazer among biped robots

Built using a $1 million grant from the Defense Advanced Research Projects Agency (DARPA), Cassie was built through the collaborative efforts of the robotics and AI departments at OSU in 16 months.

The robot is the first bipedal bot to use machine learning to control its gait in outdoor terrain and can go up and down stairs. Last August, the bipedal robot created another record when it learned to run and completed an unassisted 5k around the university campus.

"Completing a 5K was about reliability and endurance, which left open the question of, how fast can Cassie run? That led the research team to shift its focus to speed," said Devin Crowley, a graduate student associated with the project who also led the Guinness effort.

Calling the record a watershed moment in the history of bipedal robots, Jonathan Hurst, the chief technology officer at Agility Robotics, where Cassie was manufactured, said, "This may be the first bipedal robot to learn to run, but it won’t be the last. I believe control approaches like this are going to be a huge part of the future of robotics. Using learned policies for robot control is a very new field, and this 100-meter dash is showing better performance than other control methods. I think progress is going to accelerate from here."