Lightwheel AI has put a fleet of L4-class Robovan driverless logistics vehicles into trial operation in the Shahe garment wholesale district of Tianhe, Guangzhou, according to Chinese outlet D1EV. The project is the company's first scaled unmanned delivery deployment in southern China and its first commercial Robovan rollout under a "capacity as a service" model, part of a broader "mass production equals operation" strategy.
The Shahe district is one of China's major garment wholesale hubs, with heavy cargo turnover. Goods there have traditionally been moved by modified electric flatbed carts, which often cause congestion during peak seasons. The area poses a tough challenge for autonomous driving: roads are complex, with heavy mixing of pedestrians, electric bikes and vehicles, some streets lack lane markings, there are many security gates, and older alleyways suffer from poor satellite signal coverage. Cargo sizes and weights also vary widely.
Built on production-car data
The Robovan is a vehicle-grade, purpose-built platform whose technology has been validated using data from mass-produced passenger cars, Lightwheel AI said. It relies on physical AI, world models and reinforcement learning to achieve L4 autonomy without high-definition maps, and has already completed the full workflow from loading-point gates to unloading points.
Lightwheel AI said it is reusing data and engineering capabilities accumulated from its L2+ assisted-driving systems in the L4 unmanned logistics business, with the Robovan serving as the core vehicle for that push. The model has already operated in Wuhu, Jinhua and Guangzhou, among other cities.
The Shahe project, described as a real-world case of complex urban last-mile logistics, is expected to generate operational data that will help advance the commercialisation of driverless delivery in China.
