With the development of information technology, autonomous driving technology has been widely used in the field of intelligent mobile devices such as mobile robots. Smart mobile devices not only reduce manual labor, facilitate life, but also improve work efficiency. Lidar, as the core obstacle avoidance sensor of autonomous driving technology, has developed rapidly.
LiDAR can obtain information such as distance, direction and velocity of the target by analyzing and calculating the transmitted and received laser signals. Lidar also plays a huge role in various fields due to its fast working speed and good resolution.
LiDAR determines the distance of the object by measuring the time difference of the laser signal, and obtains the accurate three-dimensional information of the measured object by scanning horizontally or scanning the angle in the air, as well as obtaining signals from different pitch angles. Performance redundancy and high reliability meet the different needs of autonomous driving, rail transit, urban transportation, industrial testing and other fields. Due to the high frequency and short wavelength of the laser, extremely high angular and distance resolution can be obtained, which means that lidar can obtain very clear images using range Doppler imaging.
The accurate detection of lidar obstacles is an important part of the perception of unmanned vehicles, and it is also the basic guarantee for the safe driving of unmanned vehicles. Considering that when the 3D LiDAR obtains one frame of scene data, the obstacle can be regarded as a uniform linear motion, and the point cloud contour features obtained are theoretically more accurate. If the contour features of the target are analyzed and extracted, the 3D point cloud data needs to be clustered and segmented first.
The obstacle detection method based on contour features is only suitable for the situation with simple scenes and small interference, otherwise it is easy to mistakenly detect the bushes on the outside of the road as obstacles such as vehicles. However, when the vehicle is driving normally on the road, it only pays attention to the situation within the road boundary and does not consider obstacles outside the road boundary. Therefore, the obstacle detection area can be limited by using the road boundary information, clustering and eliminating the radar point cloud data in the area of interest, effectively filtering out most of the interference data, making the obstacle information extraction fast and convenient, and greatly improving the speed and accuracy of the algorithm detection.
Lidar is a high-frequency sensor device, which will scan the same area multiple times during the time of calibration data collection, so it is necessary to solve the average value of the sample data at this stage to obtain the statistical distribution law of the total number of samples, and at the same time reduce the amount of computation and improve the real-time performance.
The LiDAR calibration board can be used for LiDAR to calibrate the sensing target distance, so that LiDAR can more accurately judge the surrounding faulty objects and their movement trajectories. The reflectivity commonly used for LiDAR calibration is % and 90%, and if the calibration accuracy is relatively high, more stepped reflectivity can be customized.