Xiaomi Automobile Technology Co., Ltd., a subsidiary of Chinese consumer electronics giant Xiaomi, released on July 26 a patent entitled “Map Generation Method, Device, Controller, Server, Storage Medium and Vehicle.”
According to the patent abstract, this method collects environmental data, such as surrounding images and movement status, while a vehicle is in a parking area. By processing the surround view image data, a top view image is generated, then divided into multiple top view sub-images according to the traffic identification information in the top view image. Then, the top view sub-images are spliced and fused based on the movement position and posture data to generate an area map, and then the server is triggered to fuse the area maps sent by multiple vehicles to generate a target area map.
The patented method can reduce the cartographic cost of garage scenarios, improve cartographic efficiency and realize the real-time updating of high-precision maps.
Xiaomi Auto has also published a number of patents before, including: transparent chassis, high-precision positioning, neural networks, traffic light duration calculation, lane line recognition, model training, automatic lane change, automatic overtaking, behavior prediction and so on.
Xiaomi has made rapid progress since it announced that it would enter the automotive field. On July 28, it was reported by Chinese media that Xiaomi’s car production project will enter the offline stage of a “soft model” car in September, followed by field tests and a winter test cycle as scheduled. In addition, Xiaomi‘s car production project is almost complete except for the official issuance of new energy vehicle licenses, the testing and verification of products in winter and summer, and the establishment of marketing and service systems.