技术文档丨Apollo 6.0的融合组件分析

从开发套件出发，手把手教你如何实践Apollo自动驾驶技术，给对自动驾驶研究感兴趣的开发者带来实质帮助，加速每一位开发者的研发进程。本文将介绍Apollo 6.0的融合组件分析，主要从以下几点为大家进行详细讲解。

• 流程图 / 组织架构图

• 方法解读

• 部分重要结构体解读

希望本次分享能够给热爱自动驾驶的小伙伴带来一些帮助。

融合组件perception/onboard/compent/fusion_component.cc调用app/obstacle_multi_sensor_fusion.cc的init和process方法完成整个融合功能。

InterProc流程图

高精地图过滤障碍物逻辑流程图

1、Init方法:

  FusionComponentConfig comp_config;
  if (!GetProtoConfig(&comp_config)) {
    return false;
  }
  AINFO << "Fusion Component Configs: " << comp_config.DebugString();

  // to load component configs
  fusion_method_ = comp_config.fusion_method();   // 融合方法
  fusion_main_sensor_ = comp_config.fusion_main_sensor(); // 融合主传感器
  object_in_roi_check_ = comp_config.object_in_roi_check(); // 是否过滤障碍物
  radius_for_roi_object_check_ = comp_config.radius_for_roi_object_check(); // 读取地图的大小，用于过滤障碍物？

  // init algorithm plugin
  ACHECK(InitAlgorithmPlugin()) << "Failed to init algorithm plugin."; //初始化融合模块和地图模块
  writer_ = node_->CreateWriter<PerceptionObstacles>(
      comp_config.output_obstacles_channel_name());  // ?两个Writer可能是记录日志用
  inner_writer_ = node_->CreateWriter<SensorFrameMessage>(
      comp_config.output_viz_fused_content_channel_name()); // 这个Writer用于可视化
  return true;
}

其中InitAlgorithmPlugin方法如下所示:

bool FusionComponent::InitAlgorithmPlugin() {
  fusion_.reset(new fusion::ObstacleMultiSensorFusion());
  fusion::ObstacleMultiSensorFusionParam param;
  param.main_sensor = fusion_main_sensor_;
  param.fusion_method = fusion_method_;
  ACHECK(fusion_->Init(param)) << "Failed to init ObstacleMultiSensorFusion"; // 初始化融合模块

  if (FLAGS_obs_enable_hdmap_input && object_in_roi_check_) { // 判断（1）高精地图准备好了? 且 (2)开启障碍物过滤
    hdmap_input_ = map::HDMapInput::Instance();
    ACHECK(hdmap_input_->Init()) << "Failed to init hdmap input."; // 初始化高精地图
  }
  AINFO << "Init algorithm successfully, onboard fusion: " << fusion_method_;
  return true;
}

2、Proc函数:

bool FusionComponent::Proc(const std::shared_ptr<SensorFrameMessage>& message) {  if (message->process_stage_ == ProcessStage::SENSOR_FUSION) { // 状态机？return true;  }  std::shared_ptr<PerceptionObstacles> out_message(new (std::nothrow)                                                       PerceptionObstacles);  std::shared_ptr<SensorFrameMessage> viz_message(new (std::nothrow)                                                      SensorFrameMessage);  bool status = InternalProc(message, out_message, viz_message);  // 下面详述  if (status) {    // TODO(conver sensor id)    if (message->sensor_id_ != fusion_main_sensor_) {      //是主传感器才给writer传送数据      AINFO << "Fusion receive from " << message->sensor_id_ << "not from "            << fusion_main_sensor_ << ". Skip send.";    } else {      // Send("/apollo/perception/obstacles", out_message);      writer_->Write(out_message);                        // Write（log）输出数据和可视化数据      AINFO << "Send fusion processing output message.";      // send msg for visualization      if (FLAGS_obs_enable_visualization) {         // Send("/apollo/perception/inner/PrefusedObjects", viz_message);        inner_writer_->Write(viz_message);      }    }  }  return status;}

其中调用了InternalProc方法：

bool FusionComponent::InternalProc(
    const std::shared_ptr<SensorFrameMessage const>& in_message,
    std::shared_ptr<PerceptionObstacles> out_message,
    std::shared_ptr<SensorFrameMessage> viz_message) {
  {
    std::unique_lock<std::mutex> lock(s_mutex_);  //锁
    s_seq_num_++;
  }

  PERF_BLOCK_START();
  const double timestamp = in_message->timestamp_;  //时间戳
  const uint64_t lidar_timestamp = in_message->lidar_timestamp_; //lidar时间戳
  std::vector<base::ObjectPtr> valid_objects;
  if (in_message->error_code_ != apollo::common::ErrorCode::OK) {  // 1 有错误时：
    if (!MsgSerializer::SerializeMsg(               // 1.1 尝试生成PerceptionObstacles物体
            timestamp, lidar_timestamp, in_message->seq_num_, valid_objects,
            in_message->error_code_, out_message.get())) {
      AERROR << "Failed to gen PerceptionObstacles object.";
      return false;
    }
    if (FLAGS_obs_enable_visualization) {          // 1.2 尝试生成可视化消息
      viz_message->process_stage_ = ProcessStage::SENSOR_FUSION;
      viz_message->error_code_ = in_message->error_code_;
    }
    AERROR << "Fusion receive message with error code, skip it.";
    return true;
  }
  base::FramePtr frame = in_message->frame_; //抽取SensorFrameMessage中的Frame数据
  frame->timestamp = in_message->timestamp_; //更新时间戳 ?为什么要在这儿更新?in_message->timestamp_貌似也不是动态变化的呀。。。

  std::vector<base::ObjectPtr> fused_objects;
  if (!fusion_->Process(frame, &fused_objects)) { // 2 调用Process函数
    AERROR << "Failed to call fusion plugin.";
    return false;
  }
  PERF_BLOCK_END_WITH_INDICATOR("fusion_process", in_message->sensor_id_); // 融合结束

  if (in_message->sensor_id_ != fusion_main_sensor_) { // 3 如果不是主传感器帧，融合后就直接返回。
    return true;
  }

  Eigen::Matrix4d sensor2world_pose =
      in_message->frame_->sensor2world_pose.matrix(); // 4 主帧的sensor2world_pose.matrix()
  if (object_in_roi_check_ && FLAGS_obs_enable_hdmap_input) { // 5 有高精地图且允许地图过滤障碍物 
    // get hdmap
    base::HdmapStructPtr hdmap(new base::HdmapStruct());
    if (hdmap_input_) {                     // 6 地图已载入
      base::PointD position;
      position.x = sensor2world_pose(0, 3);  // 6.1 主传感器在世界（地图）坐标系的位置
      position.y = sensor2world_pose(1, 3);
      position.z = sensor2world_pose(2, 3);
      hdmap_input_->GetRoiHDMapStruct(position, radius_for_roi_object_check_, //  6.2 获得主传感器周围的地图
                                      hdmap);
      // TODO(use check)
      // ObjectInRoiSlackCheck(hdmap, fused_objects, &valid_objects); // 6.3 地图检查或者过滤障碍物未实现
      valid_objects.assign(fused_objects.begin(), fused_objects.end());
    } else {
      valid_objects.assign(fused_objects.begin(), fused_objects.end());
    }
  } else {
    valid_objects.assign(fused_objects.begin(), fused_objects.end());   // 7 拷贝一份到融合数据到valid_objects
  }
  PERF_BLOCK_END_WITH_INDICATOR("fusion_roi_check", in_message->sensor_id_);// 高精地图检查结束

  // produce visualization msg
  if (FLAGS_obs_enable_visualization) {                   // 8 填充可视化数据
    viz_message->timestamp_ = in_message->timestamp_;
    viz_message->seq_num_ = in_message->seq_num_;
    viz_message->frame_ = base::FramePool::Instance().Get(); //8.1 与调度/cyber/sceduler/sceduler_factory.cc相关，暂时不管
    viz_message->frame_->sensor2world_pose =
        in_message->frame_->sensor2world_pose;
    viz_message->sensor_id_ = in_message->sensor_id_;
    viz_message->hdmap_ = in_message->hdmap_;
    viz_message->process_stage_ = ProcessStage::SENSOR_FUSION; //8.2 更新处理阶段
    viz_message->error_code_ = in_message->error_code_;
    viz_message->frame_->objects = fused_objects;              //8.3 更新融合障碍物
  }
  // produce pb output msg
  apollo::common::ErrorCode error_code = apollo::common::ErrorCode::OK;
  if (!MsgSerializer::SerializeMsg(timestamp, lidar_timestamp,          // 9 将融合结果valid_objects序列化
                                   in_message->seq_num_, valid_objects,
                                   error_code, out_message.get())) {
    AERROR << "Failed to gen PerceptionObstacles object.";
    return false;
  }
  PERF_BLOCK_END_WITH_INDICATOR("fusion_serialize_message",     // 序列化结束
                                in_message->sensor_id_);

  const double cur_time = ::apollo::cyber::Clock::NowInSeconds(); //10 获得当前时间和延时，输出时间信息和障碍物个数
  const double latency = (cur_time - timestamp) * 1e3;
  AINFO << std::setprecision(16) << "FRAME_STATISTICS:Obstacle:End:msg_time["
        << timestamp << "]:cur_time[" << cur_time << "]:cur_latency[" << latency
        << "]:obj_cnt[" << valid_objects.size() << "]";
  AINFO << "publish_number: " << valid_objects.size() << " obj";
  return true;
}

InternalProc的输入类型为：SensorFrameMessage，定义来自perception/onboard/inner_component_messages/inner_component_messages.h

class SensorFrameMessage {
 public:
  SensorFrameMessage() { type_name_ = "SensorFrameMessage"; }
  ~SensorFrameMessage() = default;
  std::string GetTypeName() { return type_name_; }
  SensorFrameMessage* New() const { return new SensorFrameMessage; }

 public:
  apollo::common::ErrorCode error_code_ = apollo::common::ErrorCode::OK; //错误码，用于功能安全

  std::string sensor_id_; //传感器id号
  double timestamp_ = 0.0; //时间戳
  uint64_t lidar_timestamp_ = 0; //?lidar UTC时间。主帧是lidar，用硬件触发，所以记录一下？
  uint32_t seq_num_ = 0; //第几帧
  std::string type_name_; //?类型名称
  base::HdmapStructConstPtr hdmap_; //高精地图

  base::FramePtr frame_; //Frame 数据 (/perception/base/frame.h)

  ProcessStage process_stage_ = ProcessStage::UNKNOWN_STAGE; // 处理阶段，枚举见下方
};

enum class ProcessStage {
  LIDAR_PREPROCESS = 0,
  LIDAR_SEGMENTATION = 1,
  LIDAR_RECOGNITION = 2,
  STEREO_CAMERA_DETECTION = 3,
  MONOCULAR_CAMERA_DETECTION = 4,
  LONG_RANGE_RADAR_DETECTION = 5,
  SHORT_RANGE_RADAR_DETECTION = 6,
  ULTRASONIC_DETECTION = 7,
  SENSOR_FUSION = 8,
  UNKNOWN_STAGE = 9,
  PROCESSSTAGE_COUNT = 10,
  LIDAR_DETECTION = 11
};

InternalProc的第一个输出参数的类型为

/apollo/modules/dreamview/frontend/src/renderer/obstacles.js

其序列化函数在/perception/onboard/msg_serializer/msg_serializer.cc

export default class PerceptionObstacles {
  constructor() {          //这里仅贴一个构造方法
    this.textRender = new Text3D();
    this.arrows = []; // for indication of direction of moving obstacles
    this.ids = []; // for obstacle id labels
    this.solidCubes = []; // for obstacles with only length/width/height
    this.dashedCubes = []; // for obstacles with only length/width/height
    this.extrusionSolidFaces = []; // for obstacles with polygon points
    this.extrusionDashedFaces = []; // for obstacles with polygon points
    this.laneMarkers = []; // for lane markers
    this.icons = [];
    this.trafficCones = []; // for traffic cone meshes
    this.v2xCubes = [];
    this.v2xSolidFaces = [];

    this.arrowIdx = 0;
    this.cubeIdx = 0;
    this.extrusionFaceIdx = 0;
    this.iconIdx = 0;
    this.trafficConeIdx = 0;
    this.v2xCubeIdx = 0;
    this.v2xSolidFaceIdx = 0;
  }

*《Apollo 6.0的融合组件分析》

https://blog.csdn.net/lcc47251/article/details/116597936

以上便是“Apollo 6.0的融合组件分析”的全部内容，如果大家对Apollo开发套件感兴趣，可以关注添加『Apollo小哥哥』（微信号：apollo_xzs）为好友，进入技术交流群，跟开发者们一起讨论哦。

©️著作权归作者所有，如需转载，请注明出处，否则将追究法律责任。