Bipedal Robot Becomes First Robot To Run 5K Race


Jack Dunhill

Social Media Coordinator and Staff Writer

clockJul 28 2021, 17:06 UTC

Cassie in action. Image Credit: Oregon State University

You can run, you can hide, but robot overlords are probably still coming for us all. Actually, you can’t really run anymore. 

In a massive leap forward for autonomous robots, Cassie the running robot has completed an entire 5-kilometer (3.1-mile) run in 53 minutes, becoming the first bipedal robot to use machine learning to control an outdoor running gait. Developed by researchers from Oregon State University and Agility Robotics, the robotic pair of legs may open doors to a host of new animal- and human-inspired machines. 


“The Dynamic Robotics Laboratory students in the OSU College of Engineering combined expertise from biomechanics and existing robot control approaches with new machine learning tools,” said Jonathan Hurst, co-founder of Agility Robotics, in a statement

“This type of holistic approach will enable animal-like levels of performance. It’s incredibly exciting." 

Cassie is actually Agility Robotics’ fourth iteration (out of six) of bipedal robots and was developed from 2009-2016. In the following two years, Cassie was available for sale (for an undisclosed price, but under $300,000) before being discontinued in 2019 to make way for Digit, their latest model. 

Cassie was designed as an open platform to develop robot movement, whilst being able to perform many everyday tasks, and this combination saw Cassie thrown into a multitude of different sectors. 


Now, despite no longer being available for sale, it appears Cassie is still well up for a challenge, and completed the run recently untethered and on a single battery. Finishing at a respectable 53 minutes and 3 seconds, Cassie would have completed it even quicker if it weren’t for some forced resets as a result of two falls (one due to an overheating computer, and one due to trying to turn too quickly). 

Watch as Cassie smashes through the 5K race. Video Credit: Oregon State University

What makes the feat so impressive is that Cassie learnt to walk by itself, using just machine-learning and a lot of failure to gain the balance and gait required to go for extended runs. Through these algorithms, researchers are accelerating the rate at which bipedal robots can traverse outdoor environments unaided, which may have huge implications in search and rescue, package delivery and much more.  

“In the not very distant future, everyone will see and interact with robots in many places in their everyday lives, robots that work alongside us and improve our quality of life,” said Hurst. 


Cassie and Digit now join the numerous other bipedal robots in development today, including Boston Dynamics’ Atlas robot, and even CalTech’s LEONARDO, which relies on thrusters to keep itself moving.  

 This Week in IFLScience

Receive our biggest science stories to your inbox weekly!

  • tag
  • robot,

  • artificial intelligence,

  • machine learning