An untethered robot invented at the Oregon State University (OSU) College of Engineering and manufactured by OSU spinout company, Agility Robotics, has established a Guinness World Record for the fastest 100 metres by a bipedal robot. Named Cassie, the robot set a time of 24,73 seconds at an average speed of 4 m/s, starting and finishing the sprint from a standing position, without falling. Unlike a human sprinter, Cassie has bird-type legs with knees that bend backwards. To learn how to sprint, the OSU researchers say the robot’s programming was trained in a week-long simulation that compressed a year’s worth of training experiences by computing numerous calculations simultaneously.
Cassie was developed as a commercial robot with a $1 million grant from DARPA, and has been used by top universities and robotics laboratories in the US as a platform for exploring machine learning. It isn’t the fastest legged robot, but the fastest bipedal. This design has particular advantages in allowing robots to traverse spaces designed for humans. Cassie’s creators say that running wasn’t the hardest part of the challenge, but getting the robot to start and stop. “Starting and stopping in a standing position are more difficult than the running part, similar to how taking off and landing are harder than actually flying a plane,” says OSU professor, Alan Fern.
The feat is especially impressive considering Cassie pulled it off blind, without an onboard camera. Instead, Cassie first learned how to run through a series of sim-to-real training sessions. OSU’s sim-to-real machine learning methods have enabled Cassie to benefit from millions of parallel-processed simulations before deployment. All that preparation ensures the robot is ready for any given task, which can include many variables, both known and unknown. Cassie has already learned how to run, hop, skip and climb stairs.
With just two legs, Cassie’s functions are limited. Agility now uses the bipedal technology developed for Cassie to power its new robot Digit, which is not only capable of walking and climbing stairs, but also has rudimentary arms for picking up and carrying small packages. This next-generation version will include a torso, arms, hands, and a head. Digit will dramatically increase functionality. It will be much more humanoid in both shape and intention.
“The key point is that sim-to-real − which teaches a system to do jobs and tasks, as opposed to traditional programming − applies much more widely than legged robotics,” Fern says. “It’s about creating a simulator where you can practice doing something. It’s a learning program, where the practice of an equivalent of years of experience can take place very fast in a computer, and then allow for the task to be safely completed.”
He explains that this represents a radical departure from the notion of trying to program a set of rules to dictate a desired action. “That’s an approach that doesn’t work, and it isn’t scalable,” he says. “The key is to program computers to learn, and then figure out how to train them. One way is through simulation, although simulation will never be a perfect reflection of the real world. So we always put in random variations to make the simulations more robust.”
As for current challenges, Agility Robotics CTO, Jonathon Hurst says that the biggest hurdle for robots is mastering the ability to navigate their environment. Although many types of robots are already deployed in industry, a human operator is often needed, and the environment is custom-fit. Such an approach won’t work in homes, which are quite different. It would be cost-prohibitive to retrofit a warehouse, let alone a single-family house, specifically for a robot. So, we need robots that can adapt, understand variables, and adjust as needed.
As for where things are headed in workforce applications, Fern envisions a future where robots train to maintain balance through physical tasks that involve variable forces beyond the robot itself, such as carrying loads or pulling carts. He can imagine a team of robots at a construction site under the command of a single operator. As costs come down, he also envisions robots in homes, performing basic tasks. One application could be performing duties to enable older adults to live more independently.
While advancements in the training of robots like Cassie, and soon Digit, have been awe-inspiring as of late, Fern and Hurst hope to see much more progress over the next five to ten years as artificial intelligence comes of age.
© Technews Publishing (Pty) Ltd | All Rights Reserved