NetEase Lingdong Unveils “LingJue”: A Fully Domestic Engineering Machinery Embodied Intelligence Model That Rivals Expert Craftsmen

At the World Artificial Intelligence Conference (WAIC) on July 26th, NetEase Lingdong, NetEase’s intelligent engineering machinery brand, unveiled its “Lingjue” embodied AI model for open-pit mining. This groundbreaking launch marks the world’s first embodied intelligence model specifically designed for excavator loading scenarios in open-pit mines.

This launch directly supports the National Energy Administration’s strategic goal of achieving intelligent continuous operation and unmanned transportation in large open-pit coal mines by 2025. It also provides a scalable technological blueprint for the “Guidelines for the Construction of Intelligent Coal Mines.”

Visitors at the NetEase Lingdong booth were able to experience the automated loading capabilities of an unmanned excavator operating in a mine in Inner Mongolia, thousands of miles away, through a smart cockpit. “I never thought AI could help us with such dangerous work!” exclaimed one observer, impressed by the smooth, unmanned operation of the excavator.

Technical Innovation Driving Industry Upgrade

“Lingjue,” as the world’s first “end-to-end” embodied AI model for engineering machinery, leverages multimodal data-driven autonomous learning technology, fundamentally disrupting traditional development paradigms.

NetEase Lingdong’s technical lead detailed three key breakthroughs:

1. Technological Path Innovation: Abandoning the traditional modular development and scene customization approach, “Lingjue” pioneers an end-to-end integrated model, achieving a qualitative leap in generalization performance.

2. Data “Flywheel” Driven: Training data comes directly from real-world mining operations, precisely addressing scene challenges that simulation data cannot solve.

3. Fully Domestic: Built on a self-developed domestic framework, the core algorithms and hardware chips are completely independently controlled, ensuring technological security and supply chain stability.

In the harsh environment of the Holingol Coal Mine in Inner Mongolia, “Lingjue” has already achieved 80% of the loading efficiency of a human operator. Nearly 70% of its operating time requires no human intervention, and it has been successfully adapted to severe conditions such as extreme cold, high dust levels, and multiple models of mining trucks. “This technology has made AI a ‘steel-armed comrade’ in the mine,” commented a representative from an Inner Mongolian open-pit coal mine after field testing. “The loading accuracy and consistency far exceeded expectations, opening up a new path for improving safety and efficiency in the industry.”

Building an Industrial Collaboration Ecosystem

To accelerate the widespread adoption of this technology, NetEase Lingdong announced two major initiatives: the first-ever open-sourcing of the “Lingjue” dataset, and the launch of the “2027 Industrial Collaboration Plan” for the entire industry. This plan, in collaboration with leading OEMs and open-pit coal mining companies, will promote joint R&D, co-creation of application scenarios, and standardization efforts through a technology sharing platform. The goal is to achieve unmanned operations in over 30 mines by 2027, leveraging the collective power of the ecosystem to accelerate technology adoption.

Autonomous Framework Extends the Boundaries of Intelligence

As the technological cornerstone of “Lingjue,” the end-to-end training framework – “Mechanical Intelligence Core” – uniquely integrates video learning, expert demonstrations, and reinforcement learning, empowering machines with the ability to dynamically evolve. This framework has supported the successful implementation of “Lingjue” in mining scenarios and is rapidly migrating to over 10 other scenarios, including port cargo handling, concrete mixing plants, and local coal sales, with future extensions planned for agriculture, intelligent manufacturing, and other fields, redefining the ecological boundaries of industrial intelligent entities.

A Chinese Solution Leading the Global Intelligent Wave

Facing the dual challenges of labor shortages and safety risks, NetEase Lingdong has already partnered with over 30 leading companies. The launch of the “Lingjue” model and “Mechanical Intelligence Core” marks a transition for engineering machinery from “remote control” to “fully autonomous operation.”

NetEase Lingdong will continue to transfer its globally leading game AI technology to real-world industries. Through technology openness and industrial collaboration, NetEase Lingdong aims to inject new quality productive forces into mines, ports, and other scenarios, providing an efficient and safe Chinese paradigm for global industrial intelligence.