Xpeng Defends Pure Vision Strategy as LiDAR Adoption Expands in China

Xpeng has reiterated its confidence in a LiDAR-free autonomous driving future, with company executives arguing that modern artificial intelligence and visual perception systems are now capable of replacing traditional physical sensing hardware in automobiles. The discussion came during a media briefing following the launch of the company’s new GX large six-seat SUV, which adopts the brand’s latest pure vision autonomous driving architecture similar to systems used by Tesla. The move highlights Xpeng’s continued strategic transition toward camera-based perception and end-to-end AI-driven driving systems in the competitive smart vehicle market of China.

Xpeng chairman and CEO He Xiaopeng stated that the company does not feel pressured by the increasing number of vehicles equipped with LiDAR in the Chinese market, particularly among models priced above 150,000 yuan. According to He, LiDAR still serves an important purpose in several industrial applications, but he no longer considers it essential for automotive autonomous driving systems. He emphasized that Xpeng remains committed to its vision-centric strategy despite broader industry trends favoring additional sensor hardware for advanced driver assistance and self-driving capabilities.

Xpeng Explains Technical Shift Toward Pure Vision Systems

Liu Xianming, who leads Xpeng’s general intelligence center, defended the company’s technical direction by explaining that the requirement for LiDAR depends entirely on a manufacturer’s software architecture and technology stack. Liu argued that autonomous driving performance should be measured by real-world effectiveness rather than the number of hardware sensors integrated into a vehicle. He added that Xpeng’s second-generation Vision-Language-Action system forms the foundation of its LiDAR-free strategy and is designed to process large volumes of visual information more efficiently than previous autonomous driving platforms.

According to Liu, traditional LiDAR systems face technical limitations in certain operating conditions. He explained that detecting distant objects or seeing through semi-transparent obstacles requires extremely high transmission power, which may conflict with strict automotive-grade safety requirements. Liu also noted that heavy rain, fog, and other extreme weather conditions can generate excessive noise points around vehicles using LiDAR-based systems. In comparison, high-resolution cameras are capable of delivering significantly richer visual information streams that can be utilized more effectively by modern large AI models and end-to-end driving algorithms.

Key Differences Between LiDAR and Pure Vision Approaches

Xpeng’s Transition Away From LiDAR

Xpeng was previously one of the strongest advocates of LiDAR technology in the automotive sector. In September 2021, the company launched the P5 electric sedan, becoming the first automaker globally to introduce LiDAR sensors in mass-produced vehicles. Higher variants of the sedan were equipped with dual LiDAR units, and several subsequent Xpeng models adopted the same strategy. At the time, LiDAR integration was viewed as a major competitive advantage for enhancing autonomous driving accuracy and safety.

However, the company’s position began to evolve as advances in computing power and large-scale AI models accelerated. Earlier autonomous driving systems often lacked the processing capability required to efficiently interpret massive visual datasets generated by cameras. Liu explained that this limitation has now changed due to rapid improvements in end-to-end AI models and high-performance computing systems. As a result, Xpeng believes that visual signals provide richer and more scalable inputs for modern autonomous driving architectures than traditional LiDAR hardware.

Industry attention around Xpeng’s changing strategy intensified in 2024 when reports first emerged that the automaker planned to remove LiDAR sensors from future models. Since late 2024, all newly launched Xpeng vehicles have transitioned to a Tesla-style pure vision autonomous driving solution. The GX SUV now represents the latest step in this broader strategic transformation as the company focuses on reducing hardware complexity while relying more heavily on AI-powered perception systems.

Frequently Asked Questions

Why is Xpeng moving away from LiDAR in autonomous driving?
Xpeng believes modern AI-powered vision systems can deliver strong autonomous driving performance without relying on LiDAR hardware. The company argues that high-resolution cameras provide richer visual information for end-to-end AI models and advanced driving algorithms. Executives also highlighted technical concerns related to LiDAR performance in extreme weather conditions and strict automotive safety requirements. Xpeng now considers its second-generation Vision-Language-Action system capable of handling complex driving environments using primarily visual perception technologies.

What is Xpeng’s pure vision autonomous driving system?
Xpeng’s pure vision autonomous driving system is a camera-based autonomous driving architecture that relies heavily on artificial intelligence and end-to-end large models. Instead of using LiDAR sensors, the system processes visual information collected through high-resolution cameras to interpret surrounding road conditions and driving scenarios. The company’s second-generation Vision-Language-Action platform powers this technology stack. Xpeng says the approach improves scalability, reduces hardware dependence, and supports advanced autonomous driving capabilities similar to the systems developed by Tesla.