Xpeng Secures Approval for L4 Robotaxi Trials in Guangzhou
In this article (4)
Xpeng secured municipal approval on Monday to launch Level 4 robotaxi trial operations across its home city of Guangzhou, marking a key step toward commercializing its autonomous transit services.
Announced by robotaxi unit head Candice Yuan, the qualification for intelligent connected vehicle demonstration operations allows Xpeng Yoyo to begin phased public trials across the metropolis using invitation codes.
The green light follows more than 2,000 internal test rides completed by August in Guangzhou, shifting the business from factory engineering toward live commercial validation.
Invitation Codes and Phased Rollout
Riders cannot book cars freely off the street yet. Candice Yuan, head of the robotaxi business, stated on Weibo that the unit will control early volume through invitation tokens while the operations team validates vehicle dispatch and road safety across urban districts.
Trial runs shift the project out of closed engineering evaluation into real passenger transit. Yuan noted the daily service routines will test operating economics and fleet software before the company offers standardized mobility packages to third-party fleet managers.
“Compute hardware inside each vehicle runs on four proprietary Turing artificial intelligence processors delivering 3,000 TOPS of processing power.”
Hardware Stack Without Maps or Lidar
The robotaxi runs on Xpeng’s flagship GX sport utility vehicle platform, adapted directly on the factory assembly line rather than retrofitted by third-party shops. The vehicle dispenses with lidar sensors and pre-mapped high-definition survey data, relying entirely on optical cameras coupled with machine learning vision networks.
Compute hardware inside each vehicle runs on four proprietary Turing artificial intelligence processors delivering 3,000 TOPS of processing power. Path planning and dynamic driving actions operate through the company’s second-generation Vision-Language-Action artificial intelligence model, known as VLA 2.0.
Fleet Platforms and Commercial Distribution
Rather than building a proprietary retail ride-hailing network to take on incumbents directly, Xpeng intends to act as a hardware supplier and software partner. The business model targets third-party ride-hailing aggregators across domestic and international cities, charging for physical vehicle builds, recurring autonomy software fees, and fleet revenue-sharing cuts.
That platform strategy shifts the capital burden of purchasing and housing thousands of battery-powered taxis onto existing fleet operators. Operating without expensive lidar rigs lowers unit manufacturing overhead, but putting camera-only Level 4 systems into dense southern Chinese street traffic transfers operational pressure directly onto onboard computing chips and remote management centers.
Timeline Toward Driverless Services
Guangzhou authorities awarded road testing rights for intelligent connected vehicles to the manufacturer in January, prompting the formal creation of a standalone robotaxi division two months later. Factory technicians completed the first production-line unit in May, followed by an initial internal run in July with chairman and chief executive He Xiaopeng booking and riding the car.
Engineers have since built out a centralized cloud platform capable of taking remote control of stalled vehicles in traffic. Commercial deployment plans call for integration with major ride platforms in 2027, when the company aims to pull human safety supervisors out of the driver seat entirely.