Tesla CEO Elon Musk said Wednesday that he believes the company’s goal of a humanoid robot is the most important thing it’s working on.
The electric vehicle manufacturer first announced its robot plans at its AI Day last year. Instead of showing off any sort of prototype, a person in a robot suit walked stiffly on stage and proceeded to dance, spin and pretend to jump rope. “Obviously that was not real,” Musk told the audience, after thanking the person for their robot performance.
On the company’s earnings call Wednesday, Musk said of its planned robot, “This I think has the potential to be more significant than the vehicle business over time.”
Musk and Tesla also revealed that they wouldn’t deliver the long awaited Cybertruck, Roadster or Semi this year. The next morning the stock plummeted, and while down about 6% at midday, he suggested in a tweet that the robot could remake the economy and add trillions in value.
But robotics and auto manufacturing experts caution that humanoid robots are expensive, difficult to build and unlikely to have a meaningful impact in the near term for the automaker. Even in controlled environments like factories, where robots are most effective, a significant impact is unlikely to arrive quickly. And it will be far more difficult for robots to navigate other places like a person’s home.
“History has taught us not to bet against technology and technologists,” Jim Womack, author of “The Machine that Changed the World,” which chronicled how auto manufacturing evolved in 20th century, told CNN Business. “But you can bet against timetables and the immediate impact.”
The first step of Musk’s trillion-dollar, world-changing vision is using the planned robot in Tesla’s factories.
“If we can’t find a use for it then we shouldn’t expect that others would,” Musk said.
Womack said a humanoid robot seems unlikely to be a game-changer for Tesla’s ability to compete in the industry. Tesla may first use its planned robot to move parts around its factories, Musk said Wednesday. But moving parts around a factory is unlikely to be a significant part of Tesla’s costs, according to Womack.
Tesla has yet to even reveal a prototype for its humanoid robot. Musk said Wednesday that Tesla would work this year on engineering and tooling to create a robot, which he is now calling “Optimus.”
Tesla has claimed the robot will stand 5 feet 8, weigh 125 pounds and be able to carry 45 pounds.
Robots are commonly used in auto manufacturing, such as to help with pressing metal into parts, welding the vehicle together and painting it. But humanoid robots, with two arms and legs, aren’t used. Tasks like putting a nut on a bolt are incredibly difficult for robots. (Musk himself has cautioned before that “excessive automation at Tesla was a mistake,” and “humans are underrated.”)
Sangbae Kim, an MIT professor researching robots, said that a challenge of humanoid robots like what Tesla envisions is that there are likely simpler solutions available to complete tasks.
“A humanoid robot won’t be as good at washing your dishes as your dishwasher,” he said.
Other forms of robots are also cheaper than humanoids. A four-legged robot, for example, is safer and simpler because it’s easier to balance, and their center of gravity tends to be lower.
Tesla isn’t the only automaker interested in humanoid robotics. Honda developed a series of robots known as “Asimo” for nearly 20 years before pulling the plug in 2018, and Hyundai bought Boston Dynamics in 2020. The company is best known for its viral YouTube videos highlighting the robot’s abilities.
Chris Atkeson, a Carnegie Mellon University professor who studies robots, said a humanoid robot could be useful if the robot needs legs to navigate a cluttered floor. Humanoid robots tend to be narrow too, he said, so they’re well-suited to any narrow passages.
Womack, the auto manufacturing expert, noted that Musk’s talk of robots arrived as the company announced that it wouldn’t be delivering any Cybertrucks, Semis or Roadsters this year, vehicles customers have long awaited, and which were originally said to be going into production in 2021, 2019, and 2020, respectively.
“There’s a pretty high correlation between Elon Musk announcing a new sort of Rube Goldberg activities and the lack of anything else to report that investors might actually be interested in,” said Womack, pointing to the previous announcement of “full self-driving,” and the long wait since for it to actually arrive.
“It does feel that these things are always queued up for a point where the routine, regular grinding on of the business is not very interesting.”
Tesla declined to comment.