Automatic driving visual sensor data acquisition method, device, equipment and medium

Abstract

The embodiment of the invention provides an automatic driving vision sensor data acquisition method, which comprises the following steps of: modeling according to a real scene, and constructing an initial simulation environment; setting the driving path and speed of the test vehicle in the simulation environment; setting a natural environment, a traffic flow, a signboard and an obstacle of the simulation environment; and testing and operating the test vehicle in the simulation environment after setting, and collecting test data of the visual sensor. According to the embodiment of the invention, the simulation scene editing platform is established, so that the problems of difficulty in collection of test data of the visual sensor and few diversity are solved; and three-dimensional graphic rendering is used as a support, a ray tracing principle is adopted, a user clicks a mouse to add or insert path key points, a driving path of a tested vehicle is fitted, a WYSIWYG editing mode is adopted, the use difficulty of software relative to other traditional industries is weakened, and the test efficiency is further improved.

Description

technical field [0001] The invention relates to the field of automatic driving, in particular to a method, device, equipment and medium for collecting data from a visual sensor for automatic driving. Background technique [0002] With the rapid development of the automobile industry, the research on autonomous driving technology has also developed rapidly. The division of functions of the automatic driving system can be divided into perception (environment perception and positioning), decision-making (intelligent planning and decision-making), execution (control execution ) three core systems, among which the accuracy of the perception system algorithm plays a decisive role in the subsequent decision-making and execution process. [0003] At present, most of the visual sensor test data in the perception system is collected manually, and the method of shooting video with photography and camera equipment in the real environment is time-consuming, labor-intensive, and data dive...

AI Technical Summary

Problems solved by technology

[0003] At present, most of the visual sensor test data in the perception system is collected manually, and the method of shooting video with photography and camera equipment in the real environment is time-consuming, labor-intensive, and data diversity is limited. It also requires a huge data storage space. To save the collected data, the amount of data and the diversity of data required for the perception system test are much higher than the existing manually collected data

Therefore, this manual data collection method has great limitations for visual sensor testing, and it also causes certain obstacles and troubles for the upgrade and optimization of the future perception system.

Images