How did Musk's "technological hard work": Optimus Prime, from PPT to reality?

 

As a Silicon Valley madman, Tesla CEO Elon Musk’s “technological fantasies”, some of which are even shocking to ordinary people, will cause strong shocks in the industry.

 

　　Once, an abandoned golf electric car ignited his love for electric cars and eventually led him to found Tesla Motors (TSLA, $265.25, $831.153 billion market cap). Today, Tesla is the world's number one electric vehicle maker and a pioneer in autonomous driving. The intelligent experience provided by Model S and Model X has greatly promoted the development of smart cars and autonomous driving around the world.

 

　　Dissatisfaction with the traffic congestion in Los Angeles gave him the idea of ​​​​building a cross-city underground Hyperloop (super high-speed rail) network, planning to build a huge underground transportation system on the east coast of the United States.

 

　　His interest in rocket launches and his curiosity about the world of space led him to found SpaceX. Today, SpaceX can not only carry passengers around the earth, but also successfully advance human exploration of the unknown universe.

 

　　This time, he appeared again with a prototype of the Tesla humanoid robot Optimus Prime.

 

　　What kind of storm will the appearance of Optimus Prime set off in the industry? Can it replicate the success of electric vehicles? Are humanoid robots ushered in the trend? Has commercialization started here?

 

　　According to Peter J. Bentley, an honorary professor of computer science at University College London and an expert in scientific media at the Royal Society, the commercialization of humanoid robots not only faces high costs, but also has a lot of room for improvement in intelligence and technology.

 

　　He told the "Daily Economic News" reporter that most robots currently have a relatively small number of drivers and sensors, and poor control and planning capabilities. Humanoid robots require a large number of actuators to ensure the flexibility of their movements.

 

　　into reality

　　On October 1, Beijing time, the 2022 Tesla Artificial Intelligence (AI) Day officially kicked off.

 

　　Under the attention of everyone, as the door of the stage installation opened, a Tesla humanoid robot prototype walked slowly to the front of the stage, waved to the audience at the scene, and performed a "lift the roof" dance. The robot code-named Optimus (Optimus), the name comes from the well-known anime "Transformers" protagonist.

Although the gait looks a little staggering and the movements are still a little stiff, the debut of the first-generation Optimus Prime marks that Tesla's humanoid robot is moving from vision to reality.

 

　　According to Musk, on the stage of AI Day, Optimus Prime walked without a tether for the first time. In addition, after Optimus Prime's brief dance, he also said that to maintain safety, they did not want the robot to do too many movements on the stage and make it "fall a dog and eat shit".

It is not difficult to see that Optimus Prime is still in a state of being "carefully protected". However, Musk emphasized that it can do more than has been shown. At the event, Tesla showed videos of Optimus Prime carrying boxes, watering plants, and moving metal rods around a car factory.

 

　　In addition, the Tesla team also showed a version of the Optimus Prime robot that is ready for production use. It is reported to contain a 2.3 kWh battery pack that runs on a Tesla SoC and supports Wi-Fi and LTE connectivity.

Musk said the Optimus Prime is expected to sell for less than $20,000 (about 140,000 yuan). He also said that he is confident that mass production will be achieved at a lower cost, and the final output will reach millions of units.

 

　　The goal, Musk said, is to make useful humanoid robots as quickly as possible. Engineers are working to reduce the power consumption and parts count of the Optimus Prime robot.

　　ambition

　　The opportunity for Musk to enter the field of humanoid robots may date back to 2018.

　　This year, Tesla's new electric car Model 3 sold well, but the production and delivery failed to keep up with the pace. In the process of expanding production capacity, the factory's automated robots dragged its feet. After several months of adjustment, it finally gradually On Tracks.

 

　　Musk admitted in an interview with CBS at the time that the Model 3 mass production plan was being held back by robots.

 

　　He gave an example at the time, saying that the Tesla factory has a "fluff robot", which is mainly responsible for placing the fiberglass pad made of fluff on the top of the Model 3 battery pack, which is naturally not a problem for ordinary people, but this kind of Robots are not very good at picking up fluff, and because of this, the product line often collapses.

 

　　Perhaps, from that point on, Musk began to have the idea of ​​making robots perform specific tasks like humans. To achieve this goal, in his view, is naturally through artificial intelligence (AI) technology.

 

　　At the first Tesla AI Day in August 2021, Musk took to the stage with his vision for a humanoid robot. There is no hand-made model and no prototype. With just a few pages of PPT, Musk's Optimus Prime is vividly described, and his blood is boiling, which shows that he has a deep heart.

 

　　In the original idea, Tesla's humanoid robot can do dangerous, heavy-duty, and boring work: it can pick up a wrench to install screws on a Tesla electric car, or it can buy food or do housework for humans, becoming a companion for humans. "good partner".

 

　　But with the gradual realization of the PPT project, Musk's ambitions for humanoid robots are no longer just to complete specific tasks in the factory, but begin to position it as the most profitable business.

 

　　Since this year, Musk has repeatedly emphasized the strategic role of robots in Tesla's future business.

 

　　In the earnings conference call in January, he said that robot development is Tesla's most important work this year, and the importance of Optimus Prime will gradually appear in the next few years, and will eventually be more important than the automotive business and FSD (Fully Self-Driving). ) is more valuable.

 

　　He also claimed that in the long run, people will eventually see Tesla as an artificial intelligence company rather than a car company or energy company.

 

　　Tesla's Texas factory even has ambitious plans to deploy thousands of "Optimus Prime" humanoid robots in the factory. The company also plans to eventually deploy millions of robots to Tesla factories around the world.

 

　　In this regard, Peter J. Bentley, an honorary professor of computer science at University College London and a science media expert at the Royal Society, commented in an email interview with the Daily Economic News reporter that as long as enough money is invested, mass production of humanoid robots is not an issue. No chance, but whether the robot can do the job as designed is another matter.

 

　　In fact, the outside world still has relatively strong confidence in Tesla's humanoid robots, which is mainly due to Tesla's own technical support.

 

　　Chris Atkeson, a professor at the School of Robotics and Human-Computer Interaction at Carnegie Mellon University, told the "Daily Economic News" reporter that Tesla's autopilot system can fully act as the brain of a humanoid robot. Atkeson is also the developer of the prototype for "Big White" in Disney's animated "Big Bang".

 

　　Soochow Securities also pointed out in a research report on the humanoid robot industry released in August that humanoid robots are similar to electric vehicles in terms of production and manufacturing. Overall, Tesla still has a comparative advantage over other manufacturers in the field of humanoid robots.

 

　　entrants

 

　　From the perspective, there is huge space in the global robot market. According to the above research report, the data shows that in 2021, the global robot market will reach 41 billion US dollars. According to data from Statista, a global comprehensive data database, the global robot sales from 2019 to 2021 will be 32.9 billion US dollars, 36.3 billion US dollars, and 41 billion US dollars respectively.

　Specifically in the field of humanoid robots, according to statistics from consulting firm SkyQuest Technology Consulting, the global humanoid robot market will be valued at US$1.48 billion in 2021. The compound annual growth rate will reach 42.1%.

 

　　In fact, before Musk entered the game, there were already many players in the field of humanoid robots.

 

　　In 1927, the world's first humanoid robot "Televox" developed by the Westinghouse Company of the United States came out. At this time, the humanoid robot could not walk, but could lift the receiver to answer the phone.

 

　　In 1972, Professor Kato Ichiro from Waseda University in Japan, known as the "Father of Humanoid Robots in the World", took the lead in solving the problem of bipedal walking of humanoid robots, and the world's first full-scale humanoid "intelligent" robot-WABOT-1 was born. .

 

　　In the 21st century, since Honda (HMC, $21.59, $38.928 billion) released the AISMO prototype, a humanoid robot capable of jumping, many technology companies and universities around the world have also launched their humanoid robots.

The more well-known ones include Boston Dynamics Atlas robot, Pepper developed by Softbank (9984, stock price 4,900 yen, market value 7.67 trillion yen), and Agility Robotics, which Amazon (AMZN, $113, market value $1.15 trillion) invested in. Digit, and Ameca of British tech company Engineered Arts. Among them, the Atlas robot is often popular because of its amazing dancing videos, and Ameca has caused a lot of discussion because of its too rich expressions.

Atlas robot image source: Boston Dynamics official website

　At the AI ​​Day scene, Musk admitted that his humanoid robot is not very perfect, and will continue to improve and refine it. He hopes that global AI talents can also join Tesla to develop more advanced robots.

 

　　Previously, Tesla has issued humanoid robot research and development personnel recruitment information, a total of about 20 positions, responsible for the design of "drive" or other key components of the robot and other work. "The high-quality code you write will run long-term on millions of humanoid robots around the world," reads one of the job postings.

 

　　This is enough to see that Musk is indeed ambitious in the field of humanoid robots.

 

　　Looking at China, companies such as Ubisoft and Iron Man Technology have also launched their own humanoid robot products.

 

　　In 2021, UBTECH will release Walker X, China's first commercialized large-scale bipedal humanoid service robot, with application scenarios including science and technology exhibition halls, film and television variety shows, commercial performances, government and enterprise exhibition halls, etc.

 

　　According to the "China Robot Industry Development Report (2022)", the scale of China's robot market is expected to reach US$17.4 billion in 2022 (US$8.7 billion for industrial robots, US$6.5 billion for service robots, and US$2.2 billion for special robots), with an average annual growth rate of five years. rate reached 22%.

 

　　With the continuous innovation and iteration of humanoid robots, its application scenarios are becoming more abundant.

 

　　"It is expected that there will be a variety of robots in the market, such as nursing robots for companionship purposes," Atkeson said. "Humanoid robots will be popular because humans are interested in both art and machines with human characteristics."

 

　　According to a report released by the McKinsey Global Institute, it is estimated that by 2030, 15% of the global workforce will be replaced by robots.

 

　　In commercial scenarios, the main application fields of humanoid robots include medical care, logistics, and agriculture. In personal scenarios, humanoid robots may undertake tasks such as housework, home security, auxiliary education, and emotional interaction.

 

　　Commercialization hurdles remain

 

　　Although the humanoid robot market seems to be a blue ocean with great potential, the commercialization exploration has not made great progress.

 

　　Take Boston Dynamics as an example. The company has always been regarded as one of the leaders in the field of humanoid robots, but it has not escaped the tragic fate of changing hands several times.

 

　　The reason is that although Atlas is one of the pioneering products in the field of humanoid robots, it has mature software and hardware systems, and has stable walking, running and jumping, handling, and obstacle avoidance and fall prevention capabilities. So there is no profit for the company. Boston Dynamics founder Marc Raibert once said: "Robot products that have not experienced at least 25 years of painstaking research are not enough for commercialization."

 

　　Boston Dynamics' woes are no exception. SoftBank's Pepper has also been discontinued, and Engineered Arts' Ameca has no clear commercial plans.

 

　　In fact, whether robots need to be made in human form is itself a subject of controversy.

 

　　Sangbae Kim, a professor of robotics research at MIT, mentioned that a challenge for humanoid robots is that there may be simpler solutions for certain tasks that do not necessarily require humanoid robots.

Image source: Photo Network-401729191

Bentley has a similar view. In Bentley's view, trying to make robots do everything humans do is not a good idea, and engineers prefer to produce specialized robots designed to perform specific functions in a more optimized way. "For example, the welding robot in the car factory is designed to complete this task, and its shape and function match." Bentley told the "Daily Economic News" reporter.

 

　　From a technical point of view, there is still a lot of room for improvement in the development of humanoid robots. Bentley told reporters that in the face of situations that have never been encountered before, it is a test of the robot's intelligence to let humanoid robots complete more complex tasks.

"Also, humanoid robots require a large number of actuators to ensure that they have sufficient degrees of freedom (flexibility of movement), these actuators are combined with a large number of sensors to detect everything in the surrounding environment, and the control system needs to control these in real time. parts so that the robot can safely perform complex tasks." Bentley further explained to reporters. "These technical barriers are difficult to break through, and most robots have a much smaller number of actuators and sensors, and poor controllability and planning capabilities."

 

　　In addition, the high cost is also a major "stumbling block" on the road to commercialization.

 

　　According to reports, the single cost of Honda ASIMO and Boston Dynamics Atlas is as high as 3 million and 1.9 million US dollars respectively, which is undoubtedly unbearable for ordinary consumers.

 

　　Finally, in Bentley's view, how to equip a light and compact power supply system with long enough battery life is also a problem that needs to be paid attention to in the development of humanoid robots. "Today's batteries have really limited runtimes."

 

　　As for the technological prospects of robots, Atkeson is relatively more optimistic.

 

　　He told the "Daily Economic News" reporter that the robot's brain is essentially an intelligent mobile phone system. Today, technology companies such as Google, Samsung, Facebook, Amazon, etc. are constantly improving smartphone systems, so the development of robot brains will also be solved in the future.

Zhang Lingxiao Every editor Lan Suying   

source: https://www.sina.com.cn/