Gasgoo Munich- The automotive industry has spanned more than a century, yet the evolution of manufacturing models has never ceased.
In 1913, Ford introduced the world’s first moving assembly line at its Highland Park plant, shifting cars from handcrafted workshops to the era of mass industrialization. Then, in 1953, Toyota introduced Just-In-Time production, reconstructing the value logic of global manufacturing with lean philosophy. These two revolutions established the fundamental paradigm of modern automotive manufacturing.
Yet, as consumer markets shift toward personalization and customization, traditional rigid assembly lines face structural bottlenecks. The industry once adhered to a rhythm of "a facelift every three years, a redesign every five," giving plants ample time to debug new models. Today, that cycle has compressed to "a facelift every six months, a redesign every a year." Chinese automakers now iterate every 1.3 years on average, with over 300 new models slated for domestic launch in 2025—roughly 3.2 new iterations daily. But production lines remain locked in old logic, requiring hours or days of shutdown for retooling, unable to adapt to the new market rhythm of mixed-model production and rapid, small-batch iteration.
At this industry turning point, Hikrobot has already accumulated extensive frontline experience in automotive manufacturing. From domestic brands like Changan, GAC, and Geely to joint ventures and foreign giants like Volkswagen and BMW, its full-stack solutions have permeated every link of the production chain—from "stamping, welding, painting, and assembly" to new energy powertrains and die-casting. It is this frontline practice that clarified the specific capabilities the industry truly needs, giving rise to the concept of "Embodied Intelligent Manufacturing."
In April 2026, Hikrobot systematically elaborated on this concept to the industry at an intelligent manufacturing conference. This was not a laboratory theory born in a vacuum, but a systematic methodology distilled from a decade of real-world practice serving over 20,000 customers across more than 1,000 niche scenarios. Its core is a dual-drive system powered by AI: "high-flexibility intelligent agents" and "replicable scenario application capabilities."
The multiplicative effect of these two elements is precisely what distinguishes embodied intelligence from traditional automation upgrades. High-flexibility intelligent agents determine the ceiling of equipment versatility, endowing machines with autonomous adaptability for multi-tasking, varied processes, and high cycle times. Replicable scenario capabilities, meanwhile, determine deployment efficiency, allowing equipment to quickly interpret fragmented scenarios and integrate stably into production flows. One solves "how well a single point can perform," the other "how far that capability can travel." Only with this dual-drive synergy can intelligent equipment truly "enter the workshop and replicate value." Next, we break down exactly how this philosophy operates on real production lines.
Three Practical Origins of Embodied Intelligence
Concepts do not arise in a vacuum. A review of Hikrobot’s deep collaboration with clients like Changan Automobile, FAW Toyota, and Zeekr reveals that the core genes of embodied intelligence have long been quietly germinating on the factory floors of these plants.
Changan Digital Intelligence Factory: The Ultimate Flexibility of Line Reconstruction
Traditional assembly lines can no longer meet the demands of parallel multi-model production and high-quality switching; changing models once required months of shutdowns for modification. Changan’s solution was to shatter this rigid chain and reconstruct it into 13 independent, parallel "islands." Each island completes a set of processes, and the connection between islands no longer relies on the physical sequence of conveyor chains. Instead, 687 Hikrobot AMRs respond dynamically to real-time production beats. Divided into latent, towing, and heavy-duty types, these AMRs collaborate to cover the entire manufacturing flow, from stamping, welding, and final assembly to die-casting and battery production.
The critical change was not simply replacing conveyor chains with AMRs, but redefining the logistics relationship between workstations: shifting from physical "hard connections" to algorithm-driven "soft scheduling." Introducing new models no longer requires halting production for modifications; it only demands adjustments to software scheduling strategies and tooling within the islands, enabling rapid launch and production. Production capacity is no longer a constant fixed at the moment the line is built, but a variable that can be dynamically reorganized according to orders.
Regarding scenario understanding and standardized implementation, Hikrobot and Changan engaged in deep co-creation, transforming decades of industry Know-How in core processes like stamping, welding, and assembly into digital workflows executable by intelligent agents.
In the stamping workshop, 3D vision guidance enables automatic recognition and boxing of non-precision-positioned containers. The system monitors container deformation in real-time, and paired with the RobotPilot platform, precisely corrects deviations for stable picking and placing. In the welding workshop, DLS series cameras capture high-precision 3D point cloud data, which RobotPilot rapidly models to allow robots to autonomously identify workpieces and complete flexible loading. Simultaneously, weld spot detection ensures body welding quality, preventing missed, duplicate, or offset welds. In the final assembly workshop, Hikrobot’s matrix detection system, combined with an automotive-specific inspection platform, achieves comprehensive, blind-spot-free detection of chassis, wiring harnesses, exteriors, and interiors—ensuring no defective product leaves the shop floor.
The two sides jointly established the automotive industry’s first Industrial AI Joint Innovation Center, solidifying these customized capabilities into standardized solutions for large-scale promotion. This achieved a weld spot group detection accuracy of 99.4%, comprehensive cost reductions exceeding 90 million yuan per plant, a 20% boost in manufacturing efficiency, and a 20% reduction in line costs.
FAW Toyota: Collaborative Intelligence at Ultra-Large Scale
If the Changan Digital Intelligence Factory verified the limits of flexibility for a single line, FAW Toyota’s rapid digital logistics upgrade validated the maturity of ultra-large-scale cluster scheduling.
Since 2019, FAW Toyota has introduced Hikrobot AMRs across multiple production bases, deploying over 2,000 units in total. It marked the first time in the automotive industry that a single factory deployed more than 1,000 AMRs, successfully replacing manually operated forklifts and tugs within the plant.
Specifically, the two parties jointly built a best-practice model for Toyota’s smart in-plant logistics across all scenarios. This achieved full-scenario coverage from inbound receiving, storage, and line feeding to empty returns and exception handling. It encompassed core scenarios such as visual automatic receiving, robot-vision integration, AMR smart warehousing, and unmanned delivery for both large and small parts, forming a closed-loop smart logistics system.
This was not merely quantitative growth, but a qualitative leap. The project achieved the industry’s first seamless AMR switching across maps and ultra-large-scale cluster scheduling, enabling robots dispersed across different workshops and floors to collaborate efficiently as a unified whole. Post-innovation, workshop capacity jumped 50%, overall efficiency rose 20%, and small-parts sorting efficiency surged 40%. This comprehensive improvement in the precision, timeliness, and safety of parts supply helped FAW Toyota’s new energy plant rank among the world's top "star factories."
More importantly, this project provided Hikrobot with valuable engineering experience in large-scale scenarios. Hikrobot abstracted these field-verified experiences into standardized, industry-agnostic capability modules, allowing more factories to build world-class smart logistics systems without starting from scratch.
Zeekr: A Benchmark for Flexible NEV Manufacturing
In the high-end new energy sector representing the future of auto manufacturing, Zeekr’s smart factory offers a benchmark sample from another dimension, precisely tailored to the core demands of fast iteration and complex processes in NEV models.
Spanning approximately 1,946 mu, Zeekr’s smart factory fully integrates 5G and digital technologies. Hikrobot helped build a "transparent, agile, intelligent" production system. In the three core workshops of final assembly, welding, and painting, over 300 AMRs replaced traditional tractor delivery modes. This reduced labor costs by 28% and boosted production efficiency by over 20%, achieving a dual optimization of efficiency and cost.
Facing a site with over 40 types of material racks ranging from 1,000mm to 4,000mm in length, two Hikrobot AMR models achieved full-scenario adaptation, resolving the pain point of insufficient adaptability in traditional logistics equipment. Simultaneously, the RCS (Robot Control System) interfaces with PLCs to achieve precise feeding at welding robot stations, facilitating the implementation of "lights-out manufacturing" and reducing energy consumption. Furthermore, the RCS flexibly plans paths based on rack dimensions and automatically avoids congestion, ensuring on-time material delivery and efficient production progress even under tight cycle times.
From Changan’s line reconstruction to FAW Toyota’s ultra-large-scale cluster collaboration, and Zeekr’s deep penetration in complex NEV scenarios, Hikrobot’s practices with automakers across different camps prove a core fact: Embodied intelligence is not a floating concept, but a new manufacturing paradigm that can be deployed on real lines, quantified in value, and replicated in solutions.
Hikrobot’s Foundation of Confidence
The ability of these three benchmark factories to achieve such cutting-edge practices is by no means a simple superposition of point technologies, but relies on the systematic, hardcore capabilities Hikrobot has built across three dimensions: proprietary technology, engineering verification, and ecosystem construction.
Driven by its three core businesses—machine vision, mobile robotics, and articulated robotics—Hikrobot achieves full coverage of the automotive line’s "perception-transport-operation" workflow. For automakers, the core value of this system lies in deep collaboration. It is not a patchwork integration of equipment from various suppliers, but a standardized, unified deployment based on a single architecture: unified hardware interface specifications, fully interconnected software protocols, and centralized control of the scheduling platform. This fundamentally reduces the integration costs and project delivery risks associated with heterogeneous systems.
Even more critical, Hikrobot’s software platform remains open and unified, supporting automakers in independent secondary development. This elevates manufacturing capability from a one-time project delivery to a long-term, operable core asset. To date, Hikrobot’s proprietary industrial software has over 600,000 licensed users, mobile robot rollouts have exceeded 180,000 units, and machine vision product shipments have surpassed 10 million. In the domestic market, one out of every two industrial cameras and one out of every three AMRs comes from Hikrobot.
The complexity of industrial scenarios dictates a vast difference between lab theory and real-world line deployment—variables like line vibration, dust, temperature fluctuations, and electromagnetic interference are difficult to simulate in a lab. Hikrobot’s approach to solving this puzzle is to be its own first customer.
In Tonglu, Hangzhou, a smart manufacturing base with a planned annual output value exceeding 10 billion yuan has officially entered production. At full capacity, it can produce 75,000 mobile robots, 5 million machine vision products, and 20,000 articulated robots annually. This base itself serves as the first verification factory for embodied intelligence: latent robots move autonomously between floors, articulated robots paired with industrial cameras complete assembly and testing, and finished products are automatically transported to the warehouse by mobile robots—achieving fully intelligent, automated operation throughout the process.
All solutions are validated across full scenarios and tested for engineering stability on Hikrobot’s own lines before being delivered to clients. This model of "verifying before others, delivering after others" significantly reduces the technical trial-and-error risks for automakers and serves as the cornerstone of trust for Hikrobot’s solution deployment.
The delivery of a single project is not the finish line, but the starting point for large-scale replication. Ensuring that a single investment generates continuous value and that mature experience from one factory rapidly migrates to many others is key to scaling embodied intelligence. The Industrial AI Joint Innovation Center built by Hikrobot and Changan has solidified customized capabilities like visual inspection and AI quality control into standardized solutions for large-scale promotion. By abstracting single-project practices into a reusable scenario process knowledge base, core capabilities such as process parameters, logistics strategies, and quality models can be rapidly replicated and migrated between factories.
On the ecosystem front, Hikrobot has gathered over 500 partners, trained more than 17,000 individuals, and seen over 6,000 engineers pass professional certification. It has led or participated in formulating 19 national standards and 10 industry or group standards. The core goal of ecosystem construction is to continuously lower the barrier to entry for embodied intelligence, making it accessible and usable for more factories. This systematic capability loop covering technology, engineering, and ecosystem is precisely the foundation of confidence that allows Hikrobot to drive embodied intelligence from concept to reality.
Conclusion
From Changan’s full-process closed-loop implementation to FAW Toyota’s large-scale replication, and Zeekr’s high-flexibility empowerment, three paths point in a single direction: Embodied intelligence is not mere talk on paper, but a new manufacturing paradigm being verified and quantified on production lines every day.
Making intelligent equipment "capable of entering the workshop" has already become a reality in these three factories; "replicating value" is now unfolding on more lines. Hikrobot’s goal is not to deliver isolated projects, but to make this capability a standard for more factories, transforming embodied intelligence from an exploration by a few pioneers into a universal practice of automotive manufacturing.
