1. What are the functions of vision sensors?
The vision sensor has the functions of lane recognition, obstacle detection, traffic sign and ground sign recognition, traffic light recognition, feasible space detection and so on.
(1) Lane recognition: lane recognition is the most basic information that can be perceived by visual sensors. With lane recognition function, lane keeping function of expressway can be realized.
(2) Obstacle detection: There are many kinds of obstacles, such as cars, pedestrians, bicycles, animals, etc. With the obstacle information, the driverless car can follow the car in the lane.
(3) Recognition of traffic signs and ground signs: traffic signs and ground signs can be used as road features to match with high-precision maps to assist positioning, and the map can also be updated based on these perception results.
(4) Traffic light recognition: The perception ability of traffic light status is very important for self-driving cars in urban areas.
(5) Detection of the passable space: The passable space represents the area where the driverless car can operate normally.
2. Environmental perception process based on visual sensors
Generally, it includes image acquisition, image preprocessing, image feature extraction, image pattern recognition, result transmission, etc. According to the specific object recognition and different recognition methods, the environment perception process will be slightly different.
3. Environmental perception process based on visual sensors
(1) Image acquisition: image acquisition is mainly through the camera acquisition of images, if it is analog signal, analog signal to be converted into digital signal, and the digital image in a certain format. According to the specific research object and application situation, choose the cost-effective camera.
(2) Image preprocessing: image preprocessing contains many contents, including image compression, image enhancement and restoration, image segmentation, etc., which should be selected according to the actual situation.
(3) Image feature extraction: In order to complete the recognition of the object in the image, the required features should be extracted on the basis of image segmentation, and these features should be calculated, measured and classified, so that the computer can classify and recognize the image according to the feature values.
(4) Image pattern recognition: There are many methods for image pattern recognition. From the perspective of feature objects extracted from image pattern recognition, image recognition methods can be divided into recognition technology based on shape features, recognition technology based on color features and recognition technology based on texture features, etc.
(5) Result transmission: The information identified by the environment sensing system is transmitted to other control systems of the vehicle or to other vehicles around the vehicle to complete the corresponding control functions.