by Xpeng presents its own robotaxi solution based on the GX crossover
Chinese automaker Xpeng has officially unveiled its version of a robotaxi, which is based on the GX model. Unlike some competitors, the company did not create a unique car from scratch but adapted an existing production crossover. This has significantly reduced development and production costs.
The new product has already been launched into production. Xpeng plans to use these vehicles for its own driverless taxi service.
Technical features and autonomous systems
It is not yet known whether the robotaxi uses a hybrid powertrain with a range extender, like the regular GX, or a fully electric version. However, the company has confirmed that the car is equipped with four of its own Turing AI chips, providing a computing power of 3000 TOPS. The system also includes steer-by-wire technology.
The Xpeng GX robotaxi is capable of operating at an L4 level of autonomy. Since January, the company has been testing these vehicles on public roads in China. The pilot launch of the commercial service is planned for the second half of this year.
The human factor and the rejection of LiDAR
Initially, as with the Tesla robotaxi, a human operator will sit behind the wheel of the Xpeng GX, who can intervene in an unforeseen situation. However, the Chinese company plans to eliminate such “safety officers” as early as the beginning of 2027.
Another important feature is that Xpeng, like Tesla, relies exclusively on a vision-only system. This means the car does not use LiDAR or high-definition maps but relies on cameras and an advanced artificial intelligence model.
Interior and configurations
Although there are no official photos of the robotaxi’s interior yet, it is known that it will feature tinted glass for privacy, entertainment screens for rear passengers, and new comfortable seats. The car will be offered in three interior layout options: for five, six, or seven seats.
Xpeng has already obtained permits to test L4 level autonomous systems in China.
This approach to creating a robotaxi based on a production model, rather than a separate development, could be a more pragmatic path for the mass adoption of driverless technologies. The rejection of expensive sensors like LiDAR in favor of a purely camera-based system also indicates Xpeng’s desire to make the technology more accessible. At the same time, having a human operator in the initial phase is a standard global practice that allows algorithms to be safely tested in real-world conditions. The plans to eliminate “safety officers” in just a year and a half seem ambitious, considering the complexity of traffic in Chinese megacities.
