You may recognize Cassie as the basis for Agility Robotics’ delivery bot, Digit. However, if you’ve ever followed the evolution of the technology, you undoubtedly know it came into being as the Strauss-inspired Cassie. The robot consists only of legs and not much else.
Not only did the robot fuel Agility’s commercial ambitions, but it has also proven to be a solid platform for exploring the two-legged position. Announced by Professor Jonathan Hunt at Oregon State University in 2017, Cassie was founded with the aim of a $ 1 million grant from the DoD – a pretty well-known story in the robotics world.
Today a team from the Dynamics Robotics Laboratory at the OSU College of Engineering demonstrated how they continue to push Cassie to her two-legged limits. The team says the robot was able to run 5K without wires on a single charge. Cassie isn’t set to break human world records anytime soon, but the 53-minute (and three-second) run was still an impressive show for the technicians.
The robot ran for about 6.5 minutes of troubleshooting as the team dealt with an overheated computer and a botched spin that ripped it off its feet.
“Cassie is a very efficient robot because of its design and construction, and we have really been able to push the limits of the hardware and show what it can do,” said Ph.D. said the student Jeremy Dao in a statement.
According to the team, Cassie essentially taught herself to walk using a deep reinforcement learning algorithm that enabled the system to figure out how to stay upright by shifting balance while running.
“Deep reinforcement learning is a powerful method in AI that enables skills such as running, jumping and climbing stairs,” adds Yesh Godse.
In May of this year, the OSU team also demonstrated how Cassie can go up and down stairs without lidar or on-board cameras.