1.整体架构

2.业务流程

2.1 Lidar Detect Component

输入topic: /apollo/sensor/lidar16/compensator/PointCloud2
输出topic: /perception/inner/DetectionObjects

2.2 Lidar Recognition Component

输入topic: /perception/inner/DetectionObjects
输出topic: /perception/inner/PrefusedObjects

2.3 Lidar Fusion Component

输入topic: /perception/inner/PrefusedObjects
输出topic: /perception/vehicle/obstacles
输出topic: /perception/inner/visualization/FusedObjects

2.4 V2X Component

输入topic: /perception/vehicle/obstacles
输入topic:/apollo/v2x/obstacles
输入topic:/apollo/localization/pose
输出topic: /apollo/perception/obstacles

3.检测模型启动

3.1模型的单独测试

//模型的单独测试
//pointpillars detect test
./bazel-bin/modules/perception/lidar/lib/detector/point_pillars_detection/point_pillars_test

//cnn seg test
./bazel-bin/modules/perception/lidar/lib/detector/cnn_segmentation/cnn_segmentation_test

3.2检测模型的配置

//更换检测模型配置文件
modules/perception/production/data/perception/lidar/models/lidar_obstacle_pipeline/velodyne16/lidar_obstacle_detection.conf
主要的模型：PointPillarsDetection，CNNSegmentation，NCutSegmentation

Apollo6的点云检测模型详解：https://blog.csdn.net/nh54zyt/article/details/110674472

3.3激光点云的保存pcd

//保存点云为pcd
modules/perception/lidar/tools/exporter/msg_exporter_main.cc
保存地址为当前程序运行目录