Humanoid robots are no longer just a demonstration of technology fairs and futuristic laboratories. BMW now plans to introduce them directly to the production lines of its European factories, which represents one of the most important steps towards a more massive use of AI robots in the automotive industry.
The German carmaker has confirmed that it will start using two humanoid robots from Hexagon Robotics at its Leipzig plant this summer. Robots are currently in the testing phase, but BMW openly says that it sees this kind of technology as the future of car manufacturing.
“This will be the future of automotive manufacturing,” said Michael Nikolaides, director of process management and digitization at BMW.
Industrial automation is not new to the automotive industry. Robotic arms and automated systems have been used for decades in factories around the world. However, humanoid robots represent a completely different approach. Instead of the factory adapting to the machines, now the machines are adapting to the factory.
This is precisely why BMW uses robots that have a human form. According to Nikolaides, the humanoid form allows robots to work in practically every place where a human works today, without the need for expensive changes to production lines. This is an important shift in the economics of automation. Industrial robots used to be so expensive that companies had to reorganize entire factories around them. Today, robots are significantly cheaper, while changing the production infrastructure has become more expensive and complicated.
“When a robot costs $17 million, you reorganize the factory around the robot. Today, that’s no longer the case,” explains Bill Ray, an analyst at Gartner.
BMW will use robots called Aeon. These are humanoid machines 1.65 meters tall and weighing around 60 kilograms. They can carry a load of up to 15 kilograms for a short time, or about eight kilograms continuously, while moving at a maximum speed of 2.4 meters per second. Unlike some other humanoid robots that try to imitate human gait as closely as possible, Aeon uses wheels instead of legs. BMW believes that such an approach is more practical for a factory environment because the robot can move between workstations more efficiently.
“On the factory floor it makes a lot more sense because the Aeon can just be driven from one place to another,” says Nikolaides.
The robot is equipped with 21 sensors, including cameras, radar, microphones and force sensors, allowing it to precisely manipulate objects and react to its environment. The way the robots were trained is particularly interesting. BMW and Hexagon used a combination of teleoperation and simulation in a digital factory twin developed with the help of Nvidia software. This means that humans first showed the robots how to perform tasks using motion sensors, while the AI system simultaneously simulated the same processes thousands of times to find the most efficient way to work. This approach is known as reinforcement learning.
Hexagon particularly emphasizes the importance of the so-called imitation learning approach, where the robot learns by observing human movements, either through video recordings or sensors placed on the person. According to Arnaud Robert, president of robotics at Hexagon, this method can cut robot training from several months to just a few days.
“The best transfer of knowledge from a human to a robot happens when the teacher and the student have the same form,” explains Robert.
This practically means that in the future humanoid robots could learn new tasks almost as well as humans. It would be enough for them to observe the man while he is doing the work.
“That’s the ultimate goal and we’re probably a year or two away from that,” says Robert.
At BMW, Aeon robots will for now perform relatively simple tasks such as fetching parts with tools and “pick and place” operations in battery assembly. However, the company believes that precisely these types of jobs will represent the first large-scale commercial application of humanoid robots. Special focus is placed on jobs that are repetitive or physically demanding for people. BMW also speaks openly about the labor shortage problem affecting the industry across Europe.
“We know that there will be a shortage of workers in the coming years, and humanoid robots can help,” says Nikolaides.
At the same time, the company is trying to reduce fears of job losses. BMW draws a parallel with automation from the seventies of the last century, when the first industrial robots appeared.
“Back then, everyone said that automation would destroy jobs, but the opposite happened. New jobs were created,” claims Nikolaides.
BMW isn’t the only automaker aggressively exploring humanoid robotics. Toyota plans to use Digit robots from Agility Robotics, China’s Xiaomi is already testing its own humanoid robots in the production of electric vehicles, while Hyundai is developing cooperation with Boston Dynamics, whose majority owner. BMW already has experience with humanoid robots in the US as well. The Figure O2 robot previously participated in the assembly of approximately 30,000 BMW X3 vehicles at the Spartanburg plant, working at nearly the same pace as human workers.
The company claims that AI robots have a big advantage over traditional industrial machines because they are much better able to tolerate deviations and contingencies. With classic robots, even a small movement of a metal part can cause an error and stop the process. Humanoid AI robots can analyze changes and continue working without interruption.
However, despite great optimism, some experts warn that the humanoid robot market is currently full of excessive marketing and unrealistic expectations. Gartner’s Bill Ray believes that the public often overestimates the capabilities of these systems.
“When you see a humanoid robot walking, you automatically assume it can run, jump, or climb. It can’t do any of those things, but the human brain fills in those gaps,” warns Ray.
He believes that many demonstrations of humanoid robots today are more intended to raise the value of company shares than for real industrial use. Despite this, the direction of the industry’s development seems quite clear. The combination of AI systems, advanced sensors and a humanoid form is bringing robots ever closer to environments that until now have been reserved exclusively for humans.
And, if BMW really successfully introduces humanoid robots into daily production in Europe, it could be the moment when the entire automotive industry will accelerate the transition to a completely new era of automation, writes the BBC.