Nearly 40 of Cassie’s supporters were jubilant, cheering as they crossed the finish line. Their success, they said, was a defining moment in the history of robotics. Cassie’s speed and agility, honed by AI training, have shown that bipedal robots can maneuver in taxing real-world situations while maintaining balance, a problem that has plagued designers in the past.
The race is built on Cassie’s successful 2021 completion of 5km in about 53 minutes, which showed for the first time that Cassie can stay upright for extended periods. It was also the culmination of nearly five years of work by engineering and machine learning researchers at Oregon State University and a subsidiary, Agility Robotics, paving the way for more advanced designs.
“This is the first big step for human-like robots to do real work in the real world,” said Alan Fern, a professor of artificial intelligence at Oregon State University who helped train Cassie. “Because [now]We can make robots move powerfully around the world on two legs.”
For decades, scientists, businessmen, and engineers have demanded legged robots. In the 1960s, Japanese researchers created Prototypes of two-legged machines. In the past decade, engineers at MIT and Caltech have tried to do the same. Last week, Tesla CEO Elon Musk for the first time A bipedal robot, Optimus.
The researchers said legged robots have always had problems, such as losing balance and falling.
To solve this problem, Verne teamed up with Oregon State University professor Jonathan Hunt and co-founder of Agility Robotics, to train bipedal robots using machine learning and neural networks, algorithms that mimic the way the human brain works.
The research is funded in part by the National Science Foundation and the Defense Advanced Research Projects Agency (DARPA), the secretive government organization behind innovations like the Internet.
Since 2017, the team has been training Cassie how to walk properly, using algorithms to reward the robot when it moves appropriately. “It’s all inspired by Pavlovian psychology,” Verne said. “He’s just learning to anticipate these rewards and do the right thing.”
Once the team had gotten the robot to perform well in the simulation, the next step was to see how it handled real-world environments, where surfaces are uneven, friction can change and the robot’s mass can change.
In 2021, when the team had Cassie run 5km, he learned a few things. Verne said the robot was “too stomping,” and the researchers began rewarding the robot when it moderated its gait. With a successful 100m dash this year, the team is moving to the next step: torso placement and Cassie’s header. (Agility Robotics is working on one called Digit.)
This will bring engineers one step closer to the human-like robots that could one day move around packages in warehouses, build homes or provide care for the elderly in homes, Verne said.
But such developments come with their own challenges.
Robots with heads supported on Cassie’s leg design will need peripheral vision For navigating difficult terrain. Now, he said, Cassie has to look around the world, understand what things are out there and not crash into it.
The robot would also have to identify the object as something to capture, and then be smart enough to do it the way a human would. (For example, Verne said, if a robot is asked to place boxes in a room, it should load the boxes from back to front.)
Most importantly, these robots must walk with intention. “When you’re in the real world sometimes you have to pay attention to where you’re stepping,” Verne said.
However, engineering experts said it would be an uphill climb to replace humans with robots.
Nancy J. Cook, professor of human systems engineering at Arizona State University, points out that robots are good at doing things like running or kicking a soccer ball. The hardest part is creating a machine that interacts with humans in a natural way.
“What they’re missing is really complex cognition,” Cook said. “There is still a deep understanding of humans required to interact with humans they don’t have.”
Cook also said that it is commendable that robots like Cassie are advancing the robotics industry, but that it seems unnecessary to build machines that simply replicate what humans do. It might be better, she said, to create robots that can do things that humans can’t.
“Why [do] Want to rebuild ourselves? ‘ she asked. ‘I think it’s a bit of science fiction, but other than the entertainment value – I think it’s an exaggeration.’
#Meet #Cassie #robots #Usain #Bolt