Factory production lines, smart transportation, hospital guidance services… At present, humanoid robots are leaving the laboratory and landing in diverse real-life scenes.
On June 10, He Xiaopeng, chairman of Xpeng Group, issued an internal letter announcing that he would personally “end” the direct management of the robot business.
同日，博世集团董事会主席史蒂凡·哈通直言，博世正在深度参与人形机器人这一产业变革，并加速布局自动化与机器人关键技术，预计将在该领域开拓数十亿欧元的业务空间。
Looking at the entire automobile industry, OEMs such as BYD, Xpeng, and Chery Automobile have announced their humanoid robot research and development plans. Leading domestic auto parts manufacturers such as Sanhua Intelligent Control and Tuopu Group have also invested heavily in building core robot production capabilities.
Relying on the advantages of reusable technologies such as autonomous driving sensors, electric drives, and electronic controls, the automotive industry has become a high-quality “training ground” for the development and implementation of humanoid robots.
Industry forecasts show that the humanoid robot market will experience explosive growth in the next few decades, and its size is expected to rival the automotive industry.
However, with the popularity of the track, the humanoid robot industry still faces many problems.
The entire automobile chain has entered the game
At present, many domestic vehicle companies have become the main force in the research and development of humanoid robots.
He Xiaopeng recently disclosed in an internal letter that Xiaopeng’s humanoid robot is at a critical stage of mass production delivery and will achieve mass production in the fourth quarter of this year.
In the first quarter of 2027, humanoid robots will be pushed into offline stores to carry out shopping guide work, and will enter more households in 2028.
理想汽车基座模型部门近期完成了新一轮组织调整，新增具身工程、具身交互、具身行为3个具身智能相关二级部门。
“从长远角度看，无论是工厂、门店还是用户都需要人形机器人。从市场格局来看，我们认为创业公司、我们这样步入千亿元收入的中厂和更大规模的大厂都会进入机器人赛道。”理想汽车董事长李想在今年一季度财报电话会上表示。
Recently, Li Ke, executive vice president of BYD Co., Ltd., publicly stated that the company is developing humanoid robots.
In addition, car companies such as Guangzhou Automobile Group and Chery Automobile have extended their business tentacles to humanoid robots.
Following the pace of OEMs, auto parts companies such as Bosch, Schaeffler, Valeo, Sanhua Intelligent Control, and Top Group are all deploying core robot components.
For example, relying on years of accumulation in the fields of sensors, electric drives, and industrial automation, Bosch has launched the ctrlX AUTOMATION platform, making robotic technology more modular and faster to integrate.
“Core technologies such as sophisticated sensor technology, software, and efficient electric drive not only support the development of autonomous driving, but also form an important cornerstone of modern robotics technology,” said Stefan Hartung.
Onboarding of humanoid robot employees
With the rapid iteration of embodied intelligence technology, the adaptation scenarios of humanoid robots continue to expand.
“机器人是标准劳动力，是任何希望在该领域有所作为的公司都会做的产品，它并不专属于某一类型的公司。只要业务需要人员，就一定会做机器人，唯一区别在于采用外购还是自制方式。”李想表示。
“在汽车4S（销售、零配件、售后服务、信息反馈）店做导购是我们预设的场景之一。除此之外，交警执勤辅助、医院导诊等多元公共服务场景，都是我们重点探索的方向。”奇瑞墨甲机器人相关工作人员告诉记者，机器人已在芜湖落地交通管理场景并投入试点应用，主要协助交警开展路口交通秩序巡查、违章行为识别提醒等基础工作。
In fact, the automotive supply chain intelligent manufacturing system is also transforming from automation to embodied intelligence.
For example, industry leaders such as SAIC Motor and CATL have deployed humanoid robot employees.
“上汽集团的人形机器人员工今年3月在别克至境E7电池量产线上岗，主要是承担电芯抓取、上料等核心工序。”一位上汽集团工作人员告诉记者，“在电池包制造环节，传统人工上料有带电操作风险与技能门槛，传统工业机器人
Under normal circumstances, it faces limitations such as large floor space and insufficient flexibility. Humanoid robots have anthropomorphic features such as visual perception, double-arm coordination, and force-controlled grasping. They can independently complete the identification of incoming battery cells and intelligent planning of grasping paths without relying on fixed programming and precise positioning of incoming materials. The floor space is less than 15% of traditional automatic workstations. ”
A report disclosed by the international consulting firm Roland Berger stated that from the perspective of application scenarios, robots currently mainly cover four types of demand directions, and different scenarios have obvious differences in the requirements for robots in terms of stability, interaction capabilities, autonomy and environmental adaptability.
但纵观全局，机器人产业当前仍处于需求快速增长阶段。
人形机器人最佳练兵场
“汽车主机厂做具身机器人，是很自然的产品线延伸，因为它的技术栈基本上是一样的。其实现在每辆车就是一
For each robot, the intelligent hardware, chips, digital architecture, digital systems, sensors, etc. of each vehicle can be shared.
“Although NIO will not be eager to directly enter the embodied intelligence track, NIO Chairman Li Bin still talked about the rationality of the automobile company’s deployment of robots in an interview with reporters.
“The automobile manufacturing industry has very complex process scenarios, strict accuracy requirements and large-scale application potential. This is recognized as the best ‘training ground’ for humanoid robots. About 70% of the technical reserves of automobile companies can be directly reused in the field of robots.” said the above-mentioned SAIC Group staff.
“We currently do not have accurate statistical data on the reuse of automotive industry technologies for humanoid robots, but sensors, electric drive technology, electronic controls, and thermal management systems for autonomous driving can be transferred to robot joints, power system research and development, etc.” A staff member from the Office of the Secretary to the Board of Directors of Tuopu Group told reporters.
Take the MEMS (Micro-Electro-Mechanical Systems) sensor component as an example. It can realize vehicle acceleration, deceleration, braking and obstacle avoidance in autonomous driving, and is a key supporting technology for humanoid robots.
“无论工厂还是居家场景，机器人要与周围环境实现安全、精准的交互，都必须具备敏锐的触觉感知能力。MEMS传感器是机器人精准拿取物体、灵敏响应物理接触的关键。依靠这些传感器，机器人能根据不同物体的特性精准调整握力，比如稳稳握住坚固的水杯，还能轻拿易碎的高脚杯。”一位博世的工作人员表示。
The US$4 Trillion Track is Begun
In addition to technological advantages, opening up a second growth curve is also a key motivation for car companies to invest in new tracks.
Data from the Passenger Car Market Information Joint Branch of the China Automobile Dealers Association show that in the first two months of 2026, the automobile industry’s profit margin dropped to 2.9%.
At the same time, the embodied intelligence track shows explosive growth potential.
Roland Berger predicts that by 2035, the robot market deployed by automobile OEMs may reach US$750 billion, and may further expand to US$4 trillion by 2050, approaching the scale of the automotive industry.
Regarding this positive prediction, Zhu Xichan, a professor at Tongji University, believes: “Car companies have laid out the humanoid robot track, which is a major move for them to explore a new profit track.”
Morgan Stanley predicts that in 2026, China’s humanoid robot sales will increase by 133% year-on-year to 28,000 units, and is expected to increase to 2.6 million units by 2035.
具身智能赛道商业化进程明显提速，但从实验室炫技走向大规模量产普及，行业仍面临售价、成本、技术、标准化等现实挑战。
On the one hand, core costs remain high.
High-end MEMS sensors, AI (artificial intelligence) chips and other core components used in humanoid robots are expensive, and the hardware cost of a single humanoid robot is still high.
According to Yushu Technology’s prospectus, the unit price of its humanoid robots will gradually drop from 593,400 yuan in 2023 to 166,400 yuan in 2025.
Although the price has dropped significantly, it is still difficult to promote it to the civilian market on a large scale.
On the other hand, most current humanoid robots can only complete preset simple actions. They lack independent decision-making, dynamic obstacle avoidance, and flexible operation capabilities in complex environments. The safety and stability of human-computer interaction still need to be improved.
In addition, embodied intelligence still faces challenges such as the industry standard system is not perfect and the profit model has not yet been fully implemented.
Li Xiang believes: “In the long term, to achieve large-scale commercialization and industrialization of humanoid robots, it will need to develop to a level similar to that of electric vehicles from 2010 to 2015, which will take at least three years or even longer. At this stage, there are still many technical problems that need to be overcome. There is no consensus on every detail and the next technical route, and these problems need to be solved.”
