Is Honda going to reinvent the assembly line?

Some time ago, Japanese automakers were carrying out a vigorous transformation. The theme of this transformation is to catch up with the competition with Chinese automakers by learning from China's development speed. These automakers include but are not limited to Toyota, Nissan, Honda, Mazda, Subaru, Suzuki...

Interestingly, Honda is also prepared to take another path, which is to rethink the design of future vehicles to adapt to the needs of robotic workers. In other words, in the past, Ford used assembly lines to build cars at a very fast speed. Now Honda hopes to use robots in its factories to build cars, with the goal of improving manufacturing efficiency.

However, to use robots for car manufacturing, Honda's new vehicles have to be redesigned. According to a recent report by Automotive News, Honda and Toyota are testing humanoid robots in their factories in Japan, from wiring harness installation to final inspection, hoping to compete with the next-generation AI robot wave from the United States and China.

This direction sounds far-fetched, but Honda has taken specific actions at its booth at the Japan Mobility Show in October. A four-fingered AI robotic hand can mimic human dexterity and can be remotely controlled as a virtual avatar. This robot, which looks like an anime character and has cameras installed in its eyes, can grasp threaded bolts and successfully screw them into the gearbox housing.

Toshihiro Mibe, CEO of Honda, put it plainly: "We are still far from having humanoid robots spread across the entire factory. But robots are gradually evolving, and eventually, parts will be transported by robots." He also said that Honda's automobile manufacturing and design will start to take this into account, and change the design of automobiles with this as the goal from the initial stage.

In other words, Honda is not pinning all its hopes on robots quickly learning human feel and tactile sensation, but is preparing to make future parts, fasteners and assembly sequences more suitable for robotic hands to grasp, align and screw in.

This is similar to Toyota's path. Toyota demonstrated a 7-foot-tall basketball robot named CUE7, and vigorously promoted the plan to deploy 400,000 new robots in various companies under the Toyota Group in the end, but Toyota's ELEY robot is still starting with simple tasks such as parts transportation, handling and positioning.

Redesign Parts First Before Robots Are Fully Deployed

Hajime Nakajima, an engineer in the Advanced Robotics Department of Honda R&D Co., Ltd., said that Honda is testing its wheeled Avatar robots to perform various tasks in Japanese factories, including installing wiring harnesses, a complex and time-consuming task that traditionally requires human dexterity to complete.

However, this device called "Avatar Robot" can be remotely controlled to protect humans from dangerous environments. Honda believes that Avatar Robot and robotic hand technologies are future growth businesses. This system uses AI technology to guide operators to smoothly pick up tiny objects and align gears.

In fact, Honda released the bipedal robot Asimo as early as 2000, becoming a pioneer in the robotics field. But this astronaut-like robot was retired in 2022. Honda said it will retain the name Asimo and name it AsimoOS, as the operating system for the company's future series of software-defined vehicles.

Toshihiro Mibe, CEO of Honda, compared this change to early design adjustments, such as the transition of assembly lines from chain conveyors to overhead hangers and automatic guided pallets. The point is that instead of waiting for AI to fully replicate human motor control and tactile dexterity, it is better to change the geometry of parts first to make it easier for existing robot hardware to complete grasping and assembly.

Some analysts call this practice a "bold strategic shift". Honda demonstrated automobile parts redesigned specifically for humanoid robotic arm assembly. This pragmatic approach transfers the burden of adaptation from complex neural networks back to the field of mechanical engineering.

However, Toyota's pace is more cautious. It is reported that Toyota's CUE7 humanoid basketball robot wears the black and red jersey of a professional team sponsored by Toyota and can accurately shoot three-pointers, but the company originally developed it as a test platform for technologies applied to working robots such as ELEY.

Toyota's ELEY, short for Embodied Learning Robot for Enhanced Yield, is a black-and-white robot with two arms, crab-like pincers, and a blue screen with facial sensors.

Hiroshi Bito, head of robotics research at Toyota Motor's Frontier Research Center, said that Toyota is testing the ELEY system for parts supply and positioning in Japanese factories, and his team is also testing other forms of embedded AI, including final inspection robots.

"Physical artificial intelligence is not yet mature, so we start with simple tasks," Bito said. "This is just the starting point." Toyota is developing robots that can work collaboratively with employees. ELEY uses wheels to move around the factory, and is equipped with robotic arms and grippers for handling parts, assisting in tasks such as parts transportation, handling and positioning in the vehicle production process.

It can be seen that Toyota adopts a step-by-step approach, because AI robots still need to learn to perform complex physical tasks reliably, with the goal of reducing repetitive work, improving production efficiency, and allowing employees to focus on tasks that require more judgment and experience.

In fact, there are many examples of this practice in the industry. Tesla stopped producing the Model S sedan and Model X crossover this year to use space at its California factory to produce its Optimus robot; Renault aims to have 350 humanoid robots by 2027; Mercedes-Benz aims to realize assembly line operations by 2030; Toyota will invest 1 trillion yen per year from 2028 to deploy robots, including AI-driven humanoid robot workers.

Everything Needs to Be Redesigned From Scratch

In May this year, Honda Motor released a sedan prototype described by the outside world as "extremely shocking", indicating major changes to future models such as the Accord or Civic. Sources say this is part of Honda's plan to launch 15 new alternative fuel vehicles globally by 2030.

This sedan prototype abandons the traditional three-box sedan shape, adopting a lower and wider body posture, a longer wheelbase and a smooth fastback roofline. It also integrates unique styling elements, such as the plastic cladding on the lower part of the body, which is rare in sedans, and draws design inspiration from the Honda 0 Series electric concept car.

Of course, it is possible that this show car is not similar to the newly designed model that Honda plans to launch. But at present, the average price of new cars worldwide is constantly rising. Automakers generally believe that if humanoid robots can fully take over those complex, repetitive and high-intensity labor in factories, the production efficiency and manufacturing cost of vehicles will be greatly changed.

This design innovation oriented to robot manufacturing may become one of the most effective paths to make vehicles affordable again in the future.

However, some insiders believe that the robotic hands on display at Honda and Toyota's booths are only visible demonstrations, and the real invisible and most difficult part lies in part geometry, fasteners and supply chain certification.

The report points out that for half a century, industrial automation has relied on rigid, single-purpose robotic arms fixed in safety cages. The selling point of humanoid robots is that bipedal robots can directly enter traditional factories designed for humans without expensive ground modifications.

However, training basic AI models to manipulate fragile wiring harnesses, cut internal plastic trim parts or align flexible rubber seals is far more difficult than training software agents.

For example, Honda's dedicated virtual avatar robot uses computer vision-guided multi-fingered robotic hands to perform complex assembly sequences, but its success largely depends on standardized parts designed specifically for mechanical grippers. By simplifying the shape of fasteners and clamping angles, the engineering team significantly reduced the assembly failure rate in early tests.

The key here is not that robots suddenly become as dexterous as humans, but that parts first become easier for robots to grasp, align and install stably. The hardware product portfolio must be re-evaluated from this perspective of collaborative design, because original mechanical flexibility is far less important than the compatibility between software and standardized physical components.

As vehicle architectures are specially customized for robot workers, the time required to integrate new humanoid robots into production lines will be shortened from months to days. Some analysts even judge that the manufacturing leaders who take the lead in adopting this kind of physical and architectural collaborative design today will determine the profit structure of the global automotive industry in the next decade.

But it is easier said than done. Some analysts believe that Japan's robotics technology faces challenges in catching up with the United States and China. Japan's hardware, including actuators and sensors, is of high quality, but cost control is a key factor.

"Japan's goals are good, but the key is what kind of value they can create. Can they complete complex tasks as quickly and accurately as humans? The value of AI robots depends on their versatility in performing various jobs and movements."

Honda's Avatar robot is still in the technology development stage, and its demonstration does not mean that Honda has fully introduced fully autonomous humanoid robot workers in all its automobile factories. Humanoid robots must prove that they can work safely, accurately and stably in busy factories, and their cost and maintenance requirements will also determine whether large-scale application is economically viable.

This article is from the WeChat Official Account "Auto Community" (ID: iAUTO2010), written by Yang Jing, published with authorization from 36Kr.