Baidu Maps Unveils Spatiotemporal Intelligence Layout and Related Data at 2026 World Internet Conference Wuzhen Summit

At the recently held 2026 World Internet Conference Wuzhen Summit, Xie Tian, Vice President of Baidu and head of Baidu Maps, disclosed the business positioning and related data of Baidu Maps. Xie Tian stated that Baidu Maps currently has a dual positioning: first, as a spatiotemporal intelligent travel tool for the general public; second, as an open spatiotemporal intelligence underlying foundation, opening up three types of capabilities—spatial perception, temporal reasoning, and intelligent decision-making—to industries such as automotive, public transportation, logistics delivery, and cultural tourism services. In terms of data, Baidu Maps has stored over 13 million kilometers of real lane-level road information, more than 300 million precise points of interest, receives real-time interactive feedback from over 600 million users each month, and its related technologies have been deployed on over 60 million mass-produced vehicles, with a single-day peak capacity of handling 650 billion high-precision positioning requests from the BeiDou system. Based on the above accumulation, Baidu Maps has built a spatiotemporal intelligence foundation system consisting of a data layer, a model layer, and an agent layer. The data layer is responsible for synchronously updating dynamic information of the real world; the model layer relies on MoT technology to complete unified modeling across spatiotemporal dimensions, with the STI-core module supporting path reasoning and planning functions, and the STI-diffusion module supporting dynamic evolution reasoning of scenarios; the agent layer provides services through products such as AI Co-pilot and Xiaodu Xiangxiang. In terms of spatial services, Baidu Maps' overseas service capabilities cover more than 200 countries and regions, and through cooperation with multiple automakers, it covers over one million parking lots, enabling continuous navigation from the parking space at the departure point to the parking space at the destination. In the field of road safety warnings, the system can identify road conditions such as sudden braking and abnormal parking and push reminders to vehicles behind. In May 2026, when a vehicle scraping and fire accident occurred at a tunnel exit in a southern city, the system issued approximately 13,000 reminders cumulatively to passing vehicles that day. In addition, Baidu Maps has also applied dynamic perception and trend prediction capabilities to scenarios such as predicting visitor flow at scenic areas and dynamically counting available parking spaces. To adapt to the invocation needs of AI agents, Baidu Maps has upgraded its access system from three dimensions: ecosystem, products, and tools. At the ecosystem level, it connects with major mainstream AI platforms; at the product level, it has launched natively adapted functions and zero-code skill packages for AI agents; at the tool level, it has launched the MAPYA natural language code generation tool and a direct connection channel to Baidu Maps CLI's AI programming tool. In the access scenarios that have already been implemented, developers can complete the access operation by entering two commands, shortening a development cycle that originally took 2 days to the minute level. Currently, Baidu Maps has gathered over 4 million developers.