Bipedal Robot Learns to Run and Completes 5K

Cassie completed the run around the university campus in 53 minutes.
Ameya Paleja
Cassie during its 5k run around the campusOregon State University

While companies are looking at flying drones to make deliveries in the future, the researchers at Oregon State University (OSU) are aiming to make bipedal robots that will not only make deliveries but also be able to help you around the house, especially the ones with stairs. Where most robots fail in being adaptable to existing conditions, Cassie, the OSU bipedal robot, recently learned how to run and even completed a 5k.  

Developed by Agility Robotics, a spin-out of the OSU research team, Cassie was launched in 2017. At first glance, the robot looks like a body-less, ostrich running around aimlessly. But the odd-looking artificial half-bird is actually the future in the making. Built for rugged use and extensive research, Cassie boasts a full-day battery life and offers open architecture for providing low-level controls, serves as a research platform for pioneering new control methods for legged locomotion and associated technologies. 

Over the past few years, the OSU research lab has been exploring the use of machine learning for Cassie. Since the process of running requires that the runner maintain his balance while staying in motion, the team used a deep reinforcement learning algorithm that allowed Cassie to learn how to run. Using the algorithm, the team is confident that tasks such as skipping and walking up and down the stairs can also be perfected by robots. 

Funded by the National Science Foundation and the DARPA's Machine Common Sense program, the team combined expertise in biomechanics and machine learning tools to give Cassie control on its running gait. The bipedal robot recently made history when it completed a 3.1 mile (5 km) run around the university campus in 53 minutes, with the research team in tow.  

The duration of the run also accounted for almost seven minutes of stoppage time, where Cassie was reset. Once because its computer over-heated and once because it executed a turn at high-speed. What are few dents in the metal frame, if a robot can be programmed to run. 

"Cassie is a very efficient robot because of how it has been designed and built," said Jeremy Dao, a Ph.D. student at the research lab, "We were really able to reach the limits of the hardware and show what it can do."

After Cassie, Agility Robotics's latest offering is Digit, a bipedal robot that can walk and climb stairs and has a pair of hands for basic manipulation tasks.  Jonathan Hurst, the head of the Robotics Lab at OSU and co-founder of Agility Robotics said, "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."

Add Interesting Engineering to your Google News feed.
Add Interesting Engineering to your Google News feed.
message circleSHOW COMMENT (1)chevron
Job Board