Digital China's Kelo Logistics launches 'Human-Robot Dance 5.0' warehouse solution with embodied AI robots

At the Digital China Ecosystem Partner Conference 2026 held on September 22, Cai Yinghua, President of Digital China Holdings, stated that "AI is restructuring the cost structure and organizational model of supply chains. Supply chain AI has moved past the debate of 'whether to use it' and entered a phase of differentiation in 'how deeply to use it.'" Cai emphasized that Digital China will leverage its logistics subsidiary, Jingke Logistics, which has深耕 the supply chain industry for over two decades.

At the AI infrastructure level, Cai believes the key to AI implementation lies in "industry expertise" and "practical usability," with the two factors being multiplicative in effect. He described the company's "two networks" as AI training grounds, while AI serves as the brain of these networks.

Four-Layer Architecture for AI-Driven Supply Chains

According to a report by Southern Metropolis Daily, Digital China has built a four-layer architecture to integrate AI into supply chains: human-machine collaboration, supply chain digital systems, AI control towers, and job-specific intelligent agents. During the conference, Jingke Logistics released several new products:

• "Jie Xiaoguo" – a next-generation enterprise-level supply chain intelligent agent
• "Human-Machine Collaboration 5.0" – a new-generation intelligent warehousing solution
• "Jiechu Zongheng: Integrated Warehousing and Transportation" – an ecological strategy

The company also officially launched the "Jieban Shenzhou: Supply Chain AI Ecosystem Co-Creation Alliance."

Rapid Growth in Embodied Intelligence and Semiconductor Warehousing

Founded in 2003, Jingke Logistics is the smart supply chain brand under Digital China Holdings. It originally supported Digital China's warehousing and logistics intellectual property big data operations.

China's 15th Five-Year Plan explicitly calls for building an efficient and smooth modern logistics system, improving the national logistics hub network, and promoting "single document" and "single container" systems to resolve bottlenecks between different transport modes. Currently, the ratio of total social logistics costs to GDP has fallen below 14%, with a policy target of further reducing it to approximately 13.5% by 2027. Industry experts note that internal cost-cutting has reached its limit, and further savings must come from network collaboration.

"Integrated Warehousing and Transportation" Strategy

Zhang Xiaohua, General Manager of Marketing at Digital China's Jingke Logistics, officially unveiled the "Jiechu Zongheng: Integrated Warehousing and Transportation" ecological strategy. The strategy is structured around two dimensions:

• Warehousing Network (Depth): Extending from origin warehouses to forward warehouses, penetrating third- to fifth-tier cities and industrial belt origin warehouse nodes. It provides seven empowerment systems covering branding, systems, operations, marketing, resource procurement, talent training, and promotional support. The goal is to expand from 200 to 500 warehousing partners within three years, creating a technology-driven logistics service platform capable of fulfilling 1 million orders daily.

• Transportation Network (Reach): Focusing on self-operated fulfillment scenarios, building an integrated fulfillment-oriented transportation network. It uses AI control towers for intelligent decision-making, customer layers for service depth, and resource layers for breadth. The plan is to expand from 300 to 1,000 core partners within three years.

The warehousing and transportation networks operate in tandem, providing end-to-end fulfillment services to customers while enabling warehouse-transport coordination and resource complementarity for partners.

Focus on High-Growth and High-Barrier Industries

Zhang Xiaohua told Southern Metropolis Daily that Jingke will cover national customers through a multi-tier warehouse model, with fulfillment tailored to customer needs. In first- and second-tier cities, the company primarily operates self-owned warehouses, while in third- and fourth-tier cities, it adopts an ecological franchise model. Key focus areas include IT, 3C electronics, cosmetics, and high-barrier sectors such as embodied intelligence and semiconductors.

"Semiconductors and embodied intelligence have extremely high requirements for warehousing infrastructure, covering everything from facility construction to in-warehouse operations," Zhang explained. Requirements include secondary modifications for temperature and humidity control, moisture protection, and dust-proof flooring. Semiconductor core warehouses require anti-static treatment, and transportation must use enclosed vehicles, sometimes air-cushion vehicles, to prevent tipping, tilting, and vibration. Due to high unit prices—a single vehicle load can be worth up to 20 million yuan—most logistics companies avoid this segment. Additionally, end-delivery for semiconductor products is extremely demanding, requiring highly specialized personnel who follow standard operating procedures, including unpacking, installation, shelving, testing, and powering on.

Demand for embodied intelligence product delivery is also growing rapidly. Zhang revealed that "our embodied intelligence robots are expanding from first- and second-tier cities to third- and fourth-tier cities. They are categorized into seven or eight sub-types based on scenarios, including financial services and hotel robots, each with different forms." While orders are surging, they are highly fragmented, with individual customers typically ordering only three to five units—unlike traditional distribution orders for laptops that can reach hundreds or thousands. This places higher demands on cost efficiency. After delivery, products require storage, assembly, and debugging, all of which depend on nationwide integrated warehousing and transportation coordination. "In the future, our warehousing network collaboration will also extend to third- and fourth-tier cities."

Embodied Intelligence in Warehouses: Heavy Training Needs and Human-Machine Collaboration

At the conference, Zhang Hupo, CTO of Jingke Logistics, released the new-generation intelligent warehousing solution "Human-Machine Collaboration 5.0," which incorporates embodied intelligence. Building on four previous generations launched since 2017, the solution adopts a "Cloud Brain + Terminal Brain" architecture:

• Cloud Brain: Handles wave aggregation, picking prioritization, and task scheduling
• Terminal Brain (Robots and Unmanned Vehicles): Handles recognition, obstacle avoidance, picking, and transportation

The system opens real business scenarios for co-creation with partners.

According to Southern Metropolis Daily, the terminal brain robots—developed in collaboration with Dajia Robotics—can walk and pick items in warehouses. During the 2026 World Artificial Intelligence Conference, the two companies signed a strategic cooperation agreement and completed the world's first global launch of embodied intelligence cluster autonomous collaborative picking. The product has been deployed at Jingke's shoe and apparel warehouse in Tianjin's Wuqing district. During the pilot phase, the project went live in just four weeks with six robots deployed. Through staggered picking between 7:00–9:00 AM, the robot picking share increased from 0.97% to 5.67%, and further climbed to 16.8% within five months.

"We provide the 'brain' of the entire solution, while our partners provide the 'body,'" Zhang Hupo explained. "The brain decides which tasks are suitable for robots or transport vehicles, telling them which orders and SKUs to pick first. The partner's body handles recognition, obstacle avoidance, object size identification, grasping, placing in baskets, finding the first transport point, avoiding obstacles, and delivering to the destination."

Partner Selection Criteria

Zhang Hupo outlined three key criteria for selecting embodied intelligence partners:

1. Technical capability: Can the solution be deployed in real-world scenarios, not just on paper?
2. Team quality: Can the team promptly analyze and resolve anomalies and deviations?
3. Scalability: After piloting with one industry customer, can the solution be quickly expanded to other industries?

Current Limitations and Future Outlook

While embodied intelligence is gaining momentum and accelerating deployment across scenarios, Zhang acknowledged that the technology is still in its early stages. In practice, embodied intelligence robots are selective about tasks—they perform better with bagged or soft SKU categories that can be easily picked using suction cups. "However, for items with complex structures, slippery glass products, or fragile items, we do not dispatch robots. Whether using suction cups or dexterous hands, robots still require extensive training to reduce anomaly rates." Additionally, robot movement efficiency and grasping speed currently lag behind human workers. Zhang believes these limitations will be gradually overcome as equipment, technology, and models improve.

As more logistics processes are replaced by AI machines—with robots running in warehouses and unmanned vehicles on roads—this vision is becoming reality. However, Zhang noted that in the short term, technology remains limited. "Both the application and training of unmanned vehicles still face challenges. So for us, in the medium to short term, we aim to achieve a state of human-machine collaboration and co-existence—where the heaviest tasks are handled by robots, while tasks requiring cognitive skills or anomaly handling that robots cannot yet manage are performed by humans."

(Source: Southern Metropolis Daily)