China Self-Driving 2025: Level 4 Robotaxis in 10+ Cities
China's autonomous driving industry leads globally in commercial deployment, with Level 4 robotaxis operating in over 10 cities including Beijing, Shanghai, Guangzhou, Shenzhen, and Wuhan. Baidu's Apollo Go completed over 8 million rides cumulatively by 2025, making it the world's largest robotaxi service. Pony.ai and WeRide secured national-level permits to operate fully driverless commercial services without safety operators. China's approach favors lidar-based sensor fusion over Tesla's camera-only vision, with Hesai Technology supplying 60% of global automotive lidar units. The government issued national standards for L3 highway driving, with several automakers launching L3 highway-assist features.
TL;DR
Level 4 robotaxis in 10+ Chinese cities. Baidu Apollo Go completed 8M+ rides. Pony.ai and WeRide operate fully driverless. Hesai supplies 60% of global automotive lidar. National L3 highway standards issued.
Key Insights
Baidu Apollo Go Fleet
Baidu Apollo Go operates over 500 robotaxis across 10 cities, completing 8 million rides cumulatively. In Wuhan, Apollo Go handles 50,000 daily rides across 300km of service area. The service achieved full driverless operation (no safety operator) in Beijing, Chongqing, and Shenzhen. Baidu plans to expand to 30 cities by 2027.
Pony.ai Commercial Service
Pony.ai received national-level permits for fully driverless commercial robotaxi operations in Beijing, Shanghai, Guangzhou, and Shenzhen. The company operates 300+ robotaxis and completed 5 million rides. Pony.ai's pricing dropped below regular taxi fares in Guangzhou, achieving cost parity for the first time.
Hesai Lidar Dominance
Hesai Technology supplies approximately 60% of global automotive lidar units, with its FT120 solid-state lidar priced at under $200 per unit. Major customers include Li Auto, XPeng, GAC, and international automakers. Hesai's lidar production exceeded 1 million units in 2025, making it the world's largest lidar manufacturer by volume.
L3 Highway Standards
China issued national standards for Level 3 conditional autonomous driving on highways, effective 2025. Several automakers launched L3 highway-assist features including Li Auto, XPeng, and Huawei-backed Seres. Liability rules allocate responsibility to automakers when L3 mode is engaged, providing legal clarity for consumers.
Side-by-Side Comparison
| Company | Robotaxi Fleet | Cities | Autonomous Level | Rides (Cumulative) |
|---|---|---|---|---|
| Baidu Apollo Go | 500+ | 10 | L4 (fully driverless) | 8M+ |
| Pony.ai | 300+ | 4 | L4 (fully driverless) | 5M+ |
| WeRide | 400+ | 6 | L4 (fully driverless) | 4M+ |
| AutoX | 200+ | 5 | L4 (safety operator) | 2M+ |
| DeepRoute.ai | 100+ | 3 | L4 (testing) | 0.5M+ |
| Momenta | N/A | 1 | L3 highway only | N/A |
| Huawei ADS | N/A (OEM) | Nationwide | L2+/L3 highway | 500K+ vehicles |
Frequently Asked Questions
China's self-driving regulation is more centralized and commercially progressive in certain aspects compared to the US: China has national-level standards for L3 highway driving with clear liability allocation to automakers, while the US lacks federal autonomous vehicle legislation, relying on state-by-state rules; China granted fully driverless commercial permits (Pony.ai, WeRide, Apollo Go) enabling revenue-generating operations without safety operators, while Waymo and Cruise operate under local permits with more restrictive conditions; China mandates data localization for all autonomous driving data, including sensor data and mapping information, while the US allows cross-border data transfer; China's approach combines national standards with city-level pilot zones (Beijing Yizhuang, Shenzhen, Wuhan), while US states compete independently with varying rules; and China's lidar-first approach contrasts with the US debate between lidar (Waymo, Cruise) and vision-only (Tesla) paradigms, with Chinese regulators effectively endorsing sensor fusion as the standard.