Since its establishment in 2016, Hikrobot has crossed the milestone of 200,000 mobile robots rolling off the production line. As a deep participant in China's intelligent manufacturing process, the company has achieved AI-native integration across its three product lines: machine vision, mobile robots, and articulated robots, moving towards "value symbiosis".
Facing the boom in embodied intelligence and the practical challenges of industrial-level implementation, Wu Yonghai, Vice President of Hikrobot, accepted an exclusive interview with China Electronics News, providing a clear reference from the industrial frontline for the embodied intelligence industry, which is currently at a crossroads of strategic choices.
Looking Back at a Decade: From the Laboratory to the Industrial Site, Mobile Robots Complete the "First Half"
China Electronics News: Hikrobot was established in 2016, marking exactly a decade so far. Looking back now, what changes have taken place in the robotics industry over these 10 years?
Wu Yonghai: The concept of robots is broad, and it is hard to unify the definition. When we were kids, our impression of robots mostly came from Europe, the United States, and Japan, and they were rarely seen in daily life. Over the past decade, robotics technology has gradually been implemented and applied. Previously, mobile robots were mostly confined to laboratories, but now the technology has matured, and they are being deployed in large quantities. This decade has been a process from technology to products, and then to applications.
I joined Hikvision in 2014, just as the company's robotics business was taking off. I mainly worked on mobile robots, solving indoor mobility issues, which involves mechatronic products, as well as architectural systems including systems, scheduling software, and sensors. At that time, the industry was still small, and we grew slowly together with our suppliers and customers.
Although the market size is now ten times larger than it was a decade ago, it is still not large enough overall. In the "first half," mobile robots were mainly used by large enterprises and Fortune Global 500 companies. Now, we hope small and medium-sized enterprises (SMEs) can also adopt them, entering the next phase.
China Electronics News: Robotics technology is gradually converging, and we see that Hikrobot has laid out three major product lines: "eyes, feet, and hands." Beyond technology and products, what else have you done?
Wu Yonghai: To sum up, Hikrobot's advantage lies in its balance, with no obvious shortcomings. From R&D and marketing networks to the supply chain and services, the entire system is solid. We also respond quickly to technological changes, we just don't talk about it much externally.
But to be honest, it is hard to have an absolute technological generation gap in the market. With continuous deep cultivation, everyone will eventually converge, and each company can only build its own barriers. Too many barriers are not good for customers or the industry. We now have a sense of the "ecosystem" and are investing heavily in it.
For example, on the mobile robot side, we have launched several practical and open software products in recent years, making it easy for small and medium-sized factories to get started and integrate with other brands. We also have certified engineers within our own system, training several thousand people every year, and cooperating with vocational and undergraduate colleges to cultivate robotics professionals needed by the market. We also provide strong support for regional ecosystem partners; otherwise, we wouldn't be able to serve so many customers.
Moving Towards "Value Symbiosis": Ecological Openness to Address Fragmented Demands
China Electronics News: In 2025, Hikrobot launched ecosystem products such as the PlantMirror simulation software and the EasyAMR system, explicitly proposing to drive the industry from a "price war" to a "value war." You have emphasized the concept of "moving towards value symbiosis" on multiple occasions. What is the industrial judgment behind this strategic transformation? How will ecological openness be specifically implemented?
Wu Yonghai: Now, intelligent manufacturing has entered the deep-water zone, and the industry is shifting from "equipment replacement" to "deep scenario cultivation." Standardized products are being deployed quickly, but the real tough battle lies in non-standard demands—special production lines, ultra-heavy materials, and complex assembly. A single product can no longer handle these. Engaging in a price war cannot solve non-standard problems; the real blue ocean is using an open ecosystem to address the fragmentation of demands.
We look at this from two levels.
The first level is the "rigid ecosystem," which solves the problem of "whether they can be connected," building the integrability and scalability of products and platforms. Taking mobile robots as an example, we built an open architecture and process engine as early as 2022 to achieve hybrid scheduling of robots across different forms. Now, including the open-architecture software platform, standardized hardware interfaces, controller kits, and PlantMirror simulation tools, these together build the "skeleton" of the ecosystem, ensuring that equipment from different manufacturers and robots of different forms can interconnect, be hybridly scheduled, and be rapidly deployed.
The second level is the "flexible ecosystem," which solves the problems of "whether they are willing to use it, whether they know how to use it, and whether they can continuously create value." This relies on the long-term symbiotic relationship between us and our partners. Currently, we have over 500 ecosystem partners and have trained more than 6,000 certified engineers through our engineer training and certification system. Coupled with the concept of "secondary development" and knowledge-sharing mechanisms, these are the "blood" of the ecosystem's operation, allowing technical capabilities to be understood, reused, and continuously evolved in real industry scenarios.
China Electronics News: Hikrobot recently announced that 200,000 mobile robots have rolled off the production line. In which scenarios are these mobile robots mainly applied, and what quantifiable value have they brought to customers after implementation?
Wu Yonghai: Over the years, Hikrobot has, to some extent, become a barometer of China's high-end manufacturing. The volume of mobile robots rolling off the production line and business growth largely reflect the prosperity of newly built factories. When there are more new production lines, the demand follows.
We now have over 3,000 models of mobile robots, basically covering industries such as automotive, lithium batteries, photovoltaics, 3C, as well as retail, express delivery, and pharmaceuticals.
In the Chongqing Changan Digital and Intelligent Factory project, our 687 AMRs achieve second-level distribution response under the AI scheduling system. The changeover time is compressed from 4 hours to 30 minutes, supporting Changan's NEV factory in "24-hour continuous production."
Not only in the domestic market, but we have also been expanding overseas. In recent years, our pace of going global has accelerated. Robots are going out with our customers, and have gradually landed in Southeast Asia, Europe, and Australia.
For example, our customer, the cross-border e-commerce company Winit, used our STRP high-density box storage solution in its Sydney park, fully utilizing the 10-meter ceiling height to maximize the use of three-dimensional space and achieve fully automated operations throughout the entire process.
Deepening "Robots Making Robots" to Make Embodied Intelligence Smarter with Use
China Electronics News: Hikrobot takes "intelligent integration, coordination of eyes, feet, and hands" as its core strategy, and all three product lines have achieved AI-native integration. From a technical perspective, what unique system-level advantages has the synergy of these three business lines produced? What is the current progress of AI integration for "eyes, feet, and hands" respectively?
Wu Yonghai: Our strategy is very clear: "intelligent integration, coordination of eyes, feet, and hands." The three product lines are being developed separately, but AI has been integrated into the products from the very beginning. Currently, all three lines have been successfully implemented.
In machine vision products, we have directly integrated deep learning models into the VM algorithm platform (Vision Master, Hikrobot's machine vision algorithm platform), overcoming non-standard scenarios such as complex defects and blurred characters that traditional algorithms cannot handle. Moreover, it possesses online learning capabilities, allowing the model to iterate itself online and become more accurate with more testing. The intelligent code reading product line integrated AI capabilities into compact hardware from the very beginning. Trained on a massive amount of real data, its recognition capability has been elevated to a new level. The recently released AI inference terminals and AI training servers are all aimed at scaling the implementation of industrial AI.
On the mobile robot side, the core is our independently developed RCS scheduling system, which utilizes deep learning. The task allocation layer performs multi-objective optimization, making decisions in seconds; the path planning layer introduces an LSTM time-series prediction network, predicting traffic flow based on historical data, shifting obstacle avoidance from passive response to active prediction. It can now support the collaboration of thousands of AMRs and sub-second task matching, meeting the requirements for flexibility, efficiency, and reliability.
For articulated robots, we have implemented an intelligent drive-control integrated design, combining a strong real-time system and iterative learning optimization algorithms. Motion strategies can be optimized based on tasks, making dynamic responses more flexible. The vision-control integrated architecture eliminates traditional hand-eye calibration, making it ready to use out of the box and significantly improving deployment efficiency. The Robotpilot guidance platform, based on RGB-D multimodal fusion and preset super models, integrates high-precision positioning for complex sorting and precision assembly, including real-time environmental perception and autonomous path planning, resulting in stronger flexible production capabilities.
China Electronics News: We noticed that at the ceremony for the 200,000th mobile robot rolling off the production line, the company unveiled a wheeled humanoid robot, obviously giving it a very high strategic positioning. So, on the basis of mobile robots, what kind of growth space is embodied intelligence expected to open up for industrial scenarios?
Wu Yonghai: The wheeled humanoid robot is not only a beneficiary of the 200,000 mobile robots rolling off the production line but also a pioneer for the next 200,000. The two form a continuous spectrum of capabilities:
Mobile robots are "specialized workers" proficient in material flow, having formed scale advantages in the existing market—standardized tasks such as warehouse handling and production line distribution. On the other hand, wheeled humanoid robots and quadruped robot dogs are "multi-purpose workers" integrating mobility, manipulation, and intelligence. They focus on complex scenarios with high risks, high precision, and frequent product changeovers, achieving "one machine for multiple uses and rapid adaptation" to expand the incremental market. The two are matched in echelons to form a complete and dynamic industrial labor unit.
In the future, we will continue to evolve this capability spectrum, deepen "robots making robots," and make embodied intelligence smarter through use in real tasks.
China Electronics News: In 2026, financing in the embodied intelligence track is active, but the industry also faces the dilemma that "performance-oriented intelligence is easy to achieve, while industrial-level implementation is difficult." How do you view the current industry status of embodied intelligence? How does Hikrobot's principle of "99.9% stability and reliability take priority over 99% intelligence" adapt to the industrial implementation of embodied intelligence?
Wu Yonghai: Embodied intelligence is not just simple components, but a high-barrier system integrating world models, multimodal large models, reinforcement learning, force control, and whole-body motion control. Many performances feature smooth movements and stunning interactions, which are essentially demonstrations of motion control and dynamic response capabilities. Only when motion control is solid enough and whole-body coordination is robust enough can a complete dynamic performance be presented. Powerful cognitive and decision-making capabilities must be built on a stable, reliable physical carrier with excellent motion control, which is indisputable.
The industrial manufacturing our company is deeply engaged in is a harsh environment with a fast pace, low fault tolerance, and ROI that must be clearly calculated. Therefore, our core principle has always been clear: placing 99.9% stability and reliability above 99% intelligence—all intelligent capabilities must be built on a proven reliable foundation. Based on this principle, we promote implementation through a dual path of "specialized workers" and "multi-purpose workers," dynamically recombining the two to systematically transform embodied intelligence into stable and reliable productivity.
Deepening Specialization, Growing Generalization: Adhering to Autonomous Controllability and Scenario Adaptation
China Electronics News: The debate over the "specialized route" versus the "general-purpose route" for embodied intelligence in the industry continues to heat up. Hikrobot has chosen to deeply cultivate industrial scenarios and pursue both routes simultaneously. What is the core consideration behind this choice? Will you layout the general-purpose humanoid robot field in the future?
Wu Yonghai: The premise of being omnipotent is being excellent at everything. Generalization does not grow out of thin air; it must grow from specialization. Before discussing generalization, we must first maximize the stable application of a single scenario and the perfect undertaking of a single process. This is the internal law of capability evolution.
Therefore, we choose to pursue both routes: deepening specialization to accumulate sufficient data and engineering experience in deterministic scenarios, and continuously evolving multi-purpose capabilities through technologies such as adaptive learning and cross-scenario autonomous decision-making, achieving "one machine for multiple uses and flexible adaptation."
Generalization needs to be built on the foundation of both. We have been paying attention to and evaluating general-purpose humanoid robots and will decide whether to launch them based on customer needs and market maturity. Currently, we have wheeled humanoid robots, dual-arm composite robots, and quadruped robot dogs, mainly out of our own judgment of industrial scenarios; wheeled robots have their own advantages. If the technology of humanoid robots becomes more and more mature, the applicable scenarios will also increase, such as some household or outdoor scenarios. If new technologies emerge, we will first verify them in our own factories, and then push them to the market once the technological verification is mature.
China Electronics News: From the perspective of global competition, Chinese mobile robot enterprises lead in shipment volume. Where are the key R&D investments of Hikrobot? In the international market, what are the company's differentiated competitive advantages?
Wu Yonghai: Our R&D investment has always revolved around two keywords: autonomous controllability and scenario adaptation.
In terms of autonomous controllability, we adhere to the full-stack independent research and development of core technologies. From underlying algorithms, controllers, and operating systems to upper-level application platforms such as RCS and iWMS, we have built a complete technology stack to reduce reliance on a single external technology. At the same time, facing the increasing number of non-standard projects, we continuously improve the customization and flexibility of our products, enabling both software and hardware to quickly adapt to complex scenarios. In addition, we actively participate in the formulation of domestic industrial internet and robotics-related standards, promoting the compatibility and mutual recognition of the technology system, and providing a stable and credible technology foundation for the industry.
In the international market, our differentiated advantages are reflected in two levels:
First, full-chain service capabilities. We provide end-to-end services from pre-sales consulting and solution design to deployment and operation and maintenance, coupled with a comprehensive software ecosystem to help customers quickly realize value implementation, rather than just delivering equipment.
Second, deep localization. In response to the regulations, certifications, and usage habits of overseas markets, we have established a localized service network, completing standard access in key markets in advance to ensure product compliance and timely service.
Looking to the Future: Driven by the Data Flywheel, Being the "Platform Builders"
China Electronics News: Looking ahead to the next three to five years, what profound changes do you think will take place in the embodied intelligence robot industry? What role does Hikrobot hope to play in this transformation?
Wu Yonghai: In the next three to five years, there will be two relatively obvious changes in the industry:
First, "specialized workers" will become more precise, and the ecosystem will continue to expand. Specialized equipment such as "eyes, feet, and hands" will accelerate monetization in high-ROI scenarios. Open platforms will lower the application threshold, and the deployment willingness of SMEs will significantly increase. At the same time, the domestic substitution and scaling of core components will drive down the cost of complete machines, and AMRs will accelerate penetration into general industry, SMEs, and even people's livelihood service fields.
Second, data-driven enhancement of multi-purpose capabilities will reshape value returns. The massive amount of industrial data accumulated during the operation of "specialized workers" (such as eyes, feet, and hands) (operational data such as perception, execution, and mobility / spatial environment data) not only optimizes production processes and supports predictive maintenance but also becomes the nourishment for the iterative growth of the capabilities of "multi-purpose workers" such as wheeled humanoid robots, dual-arm composite robots, and quadruped robots. The more the equipment is used, the smarter it becomes, and the service value also leaps accordingly.
In this process, we hope to become the "platform builders." In the future, we will continue to deeply cultivate manufacturing and logistics, solidifying machine vision, mobile robots, and articulated robots. At the same time, we will use AI and data integration capabilities to systematically enhance the "multi-purpose" capabilities of robots and promote the scaled implementation of "one machine for multiple uses." The ideal state is that data feeds back into algorithms, algorithms optimize production lines, and production lines accumulate data again, forming a positive flywheel.
Author | Yang Pengyue Editor | Qiu Jiangyong Art Editor | Malia Supervisor | Lian Xiaodong