Baidu debuts Apollo World simulation model for autonomous vehicle training
Baidu has launched Apollo World, a simulation system designed to accelerate the development of robotaxis by creating interactive, virtual road environments.
Baidu introduced Apollo World on November 3, a simulation model engineered to replicate interactive road conditions for training and testing autonomous driving systems. The technology is currently deployed to support the development of the company’s Apollo Go robotaxi division, with the capacity to generate 100,000 kilometers of synthetic driving data daily.
The system utilizes real-time camera and lidar data streams, allowing virtual traffic participants to react dynamically to the autonomous vehicle’s behavior. This functionality enables engineers to conduct repeated tests in challenging scenarios and simulate full-length trips across diverse environmental conditions.
The launch addresses the industry-wide challenge of reducing the time and capital required to localize autonomous technology for new regions. By shifting testing into a virtual environment, developers aim to mitigate the logistical hurdles of physical road testing while refining safety protocols.