China plans world’s largest driverless ride-hailing area

Baidu said the autonomous vehicle (AV) industry has long grappled with the “long tail” problem, in which an autonomous vehicle runs into a scenario it has not seen or experienced before. To address this problem, it has announced the industry’s first “AI big model” for autonomous driving, a pre-trained visual-language model, backed by the Baidu WenXin Big Model, which recognises thousands of objects, helping to enlarge the scope of semantic recognition.

It claims the model will enable autonomous vehicles to quickly make sense of an unseen object or vehicle, such as special vehicle such as a fire truck, or ambulance.

Tackling the data challenge

Baidu’s new closed-loop data system aims to address the exponential growth of data. It has introduced the concept of “fine purification, strong ingestion” to effectively identify and utilise data. To purify the data, the system leverages both on-board small AI models and cloud-based big AI model to achieve high-efficiency data mining and automated labelling.

The data ingestion architecture achieves automated training with its group-optimisation ability and data distribution understanding to effectively utilise data and further enhance the overall intelligence of autonomous driving.

Baidu has introduced the concept of “fine purification, strong ingestion” to effectively identify and utilise data.

Baidu is also helping to bring autonomous driving technology to advanced assisted driving products. Currently, the technology stack level enables the unification of L4 and L2+ smart driving products in terms of visual perception scheme, technical architecture, map unification, data interconnection and infrastructure sharing.

Baidu envisions a mutually beneficial relationship in which L4 will continue to provide advanced technology migration for L2+ smart driving products in urban use cases, while L2 data feedback will also help to improve L4 generalisation ability.