China's Baidu has introduced a new artificial intelligence system designed to help autonomous vehicles learn how to navigate real-world roads without relying entirely on physical testing. Called Apollo World, the platform can simulate road environments and predict how they change, giving robotaxis more opportunities to practice different driving situations.
The system is part of Baidu's efforts to improve autonomous driving technology while reducing the cost and time required to develop robotaxis.
According to Baidu, Apollo World can generate up to 100,000 kilometers of simulated driving every day. The company estimates this is equivalent to around three years of real-world driving tests conducted by an autonomous vehicle with a safety driver.
The simulation also provides real-time data using cameras and lidar. This enables the autonomous driving software to recognize the environment, understand the road, and maneuver the simulated environment by utilizing the same type of information a real car would gather.
Also Read: Waymo Toronto Launch: Driverless Robotaxis Target 2028
Baidu has already started using Apollo World to train its Apollo Go robotaxi fleet for international operations. Apollo Go is currently China's largest robotaxi service operator and competes with companies such as WeRide, Pony.ai, and Momenta Global.
The launch comes as Chinese technology and automotive companies look for opportunities beyond their domestic markets. Volkswagen's Chinese partner XPeng is also planning to introduce robotaxi operations in 2027.
Also Read: Waymo Expands Robotaxi Plans With Toyota's Electric SUV in Japan
Morgan Stanley estimates that China's robotaxi fleet could reach one million vehicles by 2035. Baidu's latest launch highlights how AI simulation could become an important part of autonomous driving development. By reducing dependence on physical road testing, companies hope to accelerate deployment while making robotaxi technology more adaptable to different markets.