As autonomous driving moves into the deep waters of mass production, maps become a key element in determining the driving experience. The traditional map mode faces challenges such as high cost, low freshness, poor flexibility, and single data source, which is difficult to meet the needs of autonomous driving from high-speed and expressways to city-level roads, which limits the improvement of intelligent driving experience. Therefore, the business model of maps needs to be upgraded to the cloud-based online service model.
On September 8, at the 2023 Tencent Global Digital Ecosystem Conference • Smart Mobility, Tencent officially released the latest intelligent driving cloud map, which is the first in the industry to realize the full cloudification of autonomous driving map data, and supports online updates at the element level and the fastest minute. Tencent Smart Driving Cloud Map is a map service platform for autonomous driving and cabin driving, which can combine real-time perception data, driving behavior experience information and car companies' own data assets to carry out multi-way mapping in the cloud, creating a "lighter, faster, more flexible, and more open" cloud-based "living map" to improve the experience of people and vehicles driving together.
Zhang Shaoyu, Vice President of Tencent Maps and Head of Tencent's Digital Twin Business
Zhang Shaoyu, Vice President of Tencent Maps and Head of Tencent's Digital Twin Business, said: "Smart Cloud Map is an operable 'living map' that provides plug-and-play online services and supports collaborative operations with customers. We hope that this set of products is open and allows more customers to build a data service system based on Zhijiao Cloud Map. ”
Tuyun integrates to create a "living map", which is lighter, faster, more flexible and more open
Compared with highways and expressways, city-level roads are more complex and road information changes faster, which not only requires map data to become more and more "light" and "fresh", but also requires the innovation of map service models and product forms. While traditional maps are often delivered in the form of offline data packages, Tencent Smart Cloud Map provides autonomous driving map data through cloud services, with two modes: "cloud-to-end" and "cloud-to-cloud".
In the "cloud-to-end" mode, in addition to providing navigation information, Tencent Smart Driving Cloud Map will also send map data changes, dynamic traffic, environmental information, and driving experience data to the vehicle, providing data in a variety of flexible ways. In the "cloud-to-cloud" mode, Tencent Smart Driving Cloud Diagram directly connects to the autonomous driving cloud of car companies, and car companies can use the data information provided by Smart Driving Cloud Map and their own data to better exert the value of their business data and ensure data security.
Under this model, Tencent Smart Drive Cloud Map has four core advantages: lighter, faster, more flexible, and more open.
Tencent's four core advantages of Smart Driving Cloud Map
This map is "lighter" and supports flexible publishing updates at the map feature level. Relying on the real-time perception of vehicle-end sensors and the differential capability of map elements, it can achieve feature-level update backhaul, record the high-confidence change points to the data update layer and publish, without waiting for the complete data production and release process to be completed, so as to improve the timeliness of updates. These elements include SD standard map elements such as vehicle signals, speed limit signs, and electronic eyes, as well as key elements such as high-definition data lane lines and ground traffic signs.
This map is "faster", and in the face of different changes in road conditions, Tencent's intelligent driving cloud map can be updated in minutes at the fastest. When the vehicle terminal perceives lane-level traffic events, bad weather, natural disasters, etc., it timely transmits the data back to the intelligent driving cloud map, and realizes real-time dynamic information services at the minute level and lane level through the dynamic ODD service.
Intelligent driving cloud map multi-layer morphology
This map is "more flexible", Tencent Smart Driving Cloud Map has a scalable multi-layer form, including basic map layers, update feature layers, ODD dynamic layers, driving experience layers, operation layers, etc., car companies can flexibly configure and manage layers according to their own needs, and build a data-driven operation platform suitable for themselves by combining with their own data layers.
At the same time, this map is also "more open", supporting the collaborative construction of role manufacturers such as car companies, and autonomous driving companies.
Learn driving behavior based on AI models to improve autonomous driving experience
Zhang Shaoyu introduced that Tencent's intelligent driving cloud map is specially designed for intelligent cars, hoping that through high-freshness map data, timely and effective environmental information and dynamic traffic information, and based on the driving experience mined by large models, the autonomous driving system can bring people a better driving experience.
Through the cooperation and co-construction of the environment experience layer and driving behavior experience layer based on the AI large model, Tencent's intelligent driving cloud map can provide richer driving experience data such as curve recommended speed, bumpy road section recommended speed, and pedestrian easy road information that traditional maps do not have, making the driving behavior of the autonomous driving system closer to real people and more sophisticated. For example, by learning the speed of a human driver on a bumpy road section or curve, the autonomous driving system can be given a recommended speed that is more in line with comfortable traffic. By excavating dangerous road sections to provide early warning and deceleration information, the safety experience of autonomous driving can be improved. By providing areas on the road where human drivers frequently change lanes, using large models to learn the driver's lane change style, it can improve the success rate of lane change and the comfort of lane change driving. In addition, the multi-source data such as traffic conditions, road working conditions, driving behavior, etc., combined with the energy consumption data information of the route, etc., can be used to mine the energy consumption of drivers of different styles on different roads, which can effectively improve the accuracy of energy consumption calculation.
When autonomous driving enters the deep water area of mass production, the service mode of maps will change. In the "Graph-Cloud Integration" model, Tencent Smart Driving Cloud Map further improves the freshness and flexibility of maps, providing users with a safer, more secure, and more comfortable intelligent driving experience.
Leifeng Net