Watch Cassie The Two-Legged Robot Run A 5K Without A Tether
Video screenshot via Agility Robotics
Cassie, a bipedal robot serving as the basis for Agility Robotics’ delivery robot Digit, has cleared a major milestone. The robot managed to run five kilometers (3.1 miles) on a single charge, all on its own without a tether.
Previously, engineers from Oregon State University were able to train Cassie to go up and down a flight of stairs without the help of cameras or LIDAR. Now, the same team has trained the robot to run using a deep reinforcement learning algorithm.
As reported by Engadget, the team said that Cassie had taught itself to stay upright while using the new technique. It had to make countless subtle adjustments throughout the run to reach the finish line.
An undergrad from the OSU Dynamic Robots Laboratory, Yesh Godse, explained: “Deep reinforcement learning is a powerful method in AI that opens up skills like running, skipping, and walking up and down the stairs.”
Cassie’s first run clocked in at 43 minutes and 49 seconds, with it crossing the OSU campus in 53:03, due to six and a half minutes of technical issues. The computer had overheated after executing a turn too quickly, causing the robot to fall.
Overall, the team seemed satisfied with the accomplishment, with member Jeremy Dao saying they were able to “reach the limits of the hardware and show what it can do.”
Intrigued? Take a look at Cassie on its run below.
[via Engadget, cover image via Agility Robotics]
Also check out these recent news
