TLDR
- Baidu launched Apollo World, an AI system that simulates real-world roads to train robotaxis
- It can generate up to 100,000 km of simulated driving each day
- Baidu says it is already training the Apollo Go robotaxi fleet
- Developing self-driving technology remains costly for the industry
- Morgan Stanley estimates China’s robotaxi fleet could reach 1 million vehicles by 2035
Baidu launched an AI system called Apollo World on Tuesday. It is built to simulate real-world environments for training robotaxis and to predict how those environments change. The launch comes as automakers race to cut the cost of developing self-driving vehicles.
Baidu said Apollo World can generate “up to 100,000 km” of simulated driving each day. That is about 62,000 miles. The company said this equals about three years of real-world testing by an autonomous vehicle with a safety driver.
The system can produce real-time data streams from several sources, such as camera and lidar. This lets self-driving systems sense and move through simulated environments directly.
Baidu said Apollo World is already being used to train the fleet of Apollo Go. The robotaxi service is owned by Baidu and plans to operate its fleet globally.
Apollo Go and Its Rivals
Apollo Go is the largest robotaxi service operator in China. It competes with WeRide, Pony.ai and Momenta Global.
Volkswagen’s Chinese partner XPeng also plans to launch robotaxi operations in 2027. Morgan Stanley has estimated that China’s robotaxi fleet could grow to 1 million vehicles by 2035.
Chinese robotaxi firms have mostly grown with asset-light strategies. They supply self-driving technology and partner with automakers, ride-hailing platforms and local fleet operators overseas. This means they do not carry the full cost of owning and running fleets.
Cost of Building the Technology
Developing the technology itself is still expensive. Frank McCleary, an expert in autonomous mobility and urban transportation at consulting firm Arthur D. Little, spoke to Reuters about the issue.
“They still have the capex of training models … and vehicles, a challenge that every autonomous vehicle company is going to face,” McCleary said.
He said the speed of a launch depends on how well a company’s technology can adapt to an environment. It must also learn how to drive in that place. McCleary added that AI has helped speed up advances in self-driving.
Companies are taking different approaches to training. Tesla is seeking to expand its Cybercab robotaxi fleet. It uses an “end-to-end neural network” trained on video of human driving to develop its self-driving technology.
Baidu released the details of Apollo World on November 3. Reuters reported the news from Beijing.
