China Autonomous Driving: Self-Driving Cars, Robotaxis, and AV Regulation
China has emerged as the world's leading market for autonomous vehicle deployment, with robotaxis operating commercially in over 50 cities by 2025. Baidu's Apollo Go, Pony.ai, and WeRide lead the commercialization race, while companies like Huawei, Xpeng, and Li Auto integrate advanced driver-assistance systems into consumer vehicles. China's regulatory framework has evolved rapidly, with dedicated autonomous driving zones, national standards for L3/L4 testing, and progressive commercialization permits that outpace Western counterparts.
TL;DR
Over 50 Chinese cities have approved autonomous driving tests or commercial operations. Baidu Apollo Go completed 8M+ robotaxi rides. China's AV market is projected to reach 300 billion RMB by 2030. L4 robotaxis operate driverless in Beijing, Wuhan, and Shenzhen. Huawei Qiankun ADS 3.0 offers L2+ capability in 15 vehicle models.
Key Insights
Robotaxi Fleet Scale
Baidu's Apollo Go completed over 8 million robotaxi rides across 10+ cities by 2025. Pony.ai operates commercial robotaxis in Guangzhou, Beijing, and Shanghai. WeRide expanded to 7 cities with mixed-fleet operations including autonomous minibuses.
Driverless Operations
China approved fully driverless (no safety driver) robotaxi operations in 10 cities including Beijing's Yizhuang district, Wuhan Economic Zone, and Shenzhen's Nanshan district. Operations cover 1,000+ km of approved roadways with progressively fewer restrictions.
Consumer AV Adoption
Approximately 15 million vehicles with L2+ autonomous driving capability were sold in China in 2025. Huawei's Qiankun ADS 3.0, Xpeng's XNGP, and Li Auto's NOA collectively cover highway, urban, and parking scenarios. Urban NOA reached 300+ cities.
Regulatory Leadership
China established autonomous driving test and operation zones in over 50 cities. National standards for L3 and L4 autonomous driving testing were formalized in 2024. Beijing, Shanghai, Guangzhou, and Shenzhen lead in progressive commercialization policies.
Side-by-Side Comparison
| Company | Level | Cities | Fleet Size | Key Achievement |
|---|---|---|---|---|
| Baidu Apollo Go | L4 | 10+ | 500+ vehicles | 8M+ rides, fully driverless |
| Pony.ai | L4 | 5 | 300+ vehicles | Nasdaq listed, commercial ops |
| WeRide | L4 | 7 | 200+ vehicles | Minibus + taxi, HK IPO filed |
| AutoX | L4 | 5 | 150+ vehicles | Fully driverless Shenzhen |
| Huawei ADS | L2+ | 300+ cities | OEM integration | 15M vehicles equipped |
| Xpeng XNGP | L2+ | 300+ cities | OEM integration | End-to-end neural network |
| Li Auto NOA | L2+ | All China | OEM integration | BEV + Transformer-based |
| Momenta | L2/L3 | OEM partners | Software only | 10+ OEM partnerships |
Frequently Asked Questions
Yes, robotaxis are generally accessible to tourists in China, though with some limitations: Baidu Apollo Go is available through the Apollo Go app (available on Chinese app stores) in cities including Beijing, Guangzhou, Shenzhen, Wuhan, Chongqing, and Chengdu; you need a Chinese phone number to register, which tourists can obtain by purchasing a local SIM card; fares are typically 10-50 RMB per ride depending on distance, and promotional free rides are frequently offered; the experience includes a safety briefing on the tablet inside the vehicle, and a remote human operator can assist if needed; fully driverless vehicles have no human safety driver but include an emergency stop button; operating hours vary by city but typically run 8AM-10PM with expanded hours in some cities; and the ride experience is generally smooth, though vehicles may drive conservatively compared to human drivers, particularly in complex intersections. The best cities to try robotaxis are Beijing's Yizhuang district (Baidu's headquarters area with the densest coverage) and Wuhan's Economic Development Zone (one of the first areas approved for fully driverless operations at scale).
China's autonomous driving regulatory framework differs from the US in several key aspects: China takes a more centralized approach with national standards set by MIIT (Ministry of Industry and Information Technology), while the US has a patchwork of state-level regulations with federal guidance from NHTSA; China approved fully driverless commercial operations faster, with driverless robotaxis operating in 10 cities by 2025, while California approved fully driverless operations but only through specific permits; China's testing framework requires gradual progression through L1-L5 levels with government-supervised evaluation at each stage, while the US allows companies to self-certify compliance; China mandates data localization (all AV driving data must be stored on domestic servers), while the US has no such requirement though state laws vary on data privacy; China has established designated AV operation zones with special traffic rules, while the US allows testing on public roads with permits; China's insurance framework for AV accidents assigns liability progressively (manufacturer takes more responsibility at higher automation levels), while US liability varies by state with ongoing legal uncertainty; and China has invested significantly more in roadside V2X (vehicle-to-everything) infrastructure, with smart traffic lights and sensors deployed in AV zones, while the US V2X deployment has been slower and less coordinated. Overall, China's approach is more structured and directive, while the US approach is more market-driven and decentralized. Both approaches have advantages: China enables faster scaling while the US encourages innovation diversity.