创建一个周期性执行的Autoware节点_autoware::common::

1.介绍

Autoware.Auto是在ROS2上开发的一个自动驾驶平台。其基本的运行单位与ROS2相同为节点。一个节点相当于一个完整的应用，大部分通过“RCLCPP_COMPONENTS_REGISTER_NODE”组件编译，循环执行。Autoware中绝大多数节点采用的是事件驱动，即订阅事件触发整个应用进行计算。此处以开始的point_cloud_filter_transform_nodes为例，将其改为周期性的时间触发机制。

2.Point_Cloud_Filter_Transform_Nodes基本结构

class POINT_CLOUD_FILTER_TRANSFORM_NODES_PUBLIC PointCloud2FilterTransformNode
  : public rclcpp::Node
{
public:
  /// \brief Parameter constructor
  /// \param node_options Additional options to control creation of the node.
  explicit PointCloud2FilterTransformNode(const rclcpp::NodeOptions & node_options);
 
protected:
  /// \brief Call distance & angle filter and then static transformer for all the points
  /// \param msg Raw point cloud
  /// \return Filtered and Transformed point cloud.
  /// \throws std::runtime_error on unexpected input contents or not enough output capacity
  const PointCloud2 & filter_and_transform(const PointCloud2 & msg);
 
  /// \brief Run main subscribe -> filter & transform -> publish loop
  void process_filtered_transformed_message(
    const PointCloud2::SharedPtr msg);
 
  template<typename PointType>
  /// \brief Check if the point is within the specified angle and radius limits
  /// \tparam PointType type with x, y, z.
  /// \param pt point with x, y, z
  /// \return True if the point is within the desired radius and angle limits. False otherwise.
  bool8_t point_not_filtered(const PointType & pt) const
  {
    return m_angle_filter(pt) && m_distance_filter(pt);
  }
 
  template<typename PointType>
  /// \brief Apply static transform to the given point
  /// \tparam PointType type with x, y, z
  /// \param pt point with x, y, z
  /// \return Transformed point
  PointType transform_point(const PointType & pt) const
  {
    PointType pt_final;
    m_static_transformer->transform(pt, pt_final);
    return pt_final;
  }
 
private:
  using AngleFilter = autoware::common::lidar_utils::AngleFilter;
  using DistanceFilter = autoware::common::lidar_utils::DistanceFilter;
  using StaticTransformer = autoware::common::lidar_utils::StaticTransformer;
  AngleFilter m_angle_filter;
  DistanceFilter m_distance_filter;
  const std::string m_input_frame_id;
  const std::string m_output_frame_id;
  std::unique_ptr<StaticTransformer> m_static_transformer;
  const std::chrono::nanoseconds m_init_timeout;
  const std::chrono::nanoseconds m_timeout;
  const typename rclcpp::Subscription<PointCloud2>::SharedPtr m_sub_ptr;
  const typename std::shared_ptr<rclcpp::Publisher<PointCloud2>> m_pub_ptr;
  const size_t m_expected_num_publishers;
  const size_t m_expected_num_subscribers;
  const std::size_t m_pcl_size;
  PointCloud2 m_filtered_transformed_msg;
};

其继承自rclcpp::Node父类。改成时间驱动首先要添加头文件

#include <chrono>
#include <functional>

其次在私有属性里添加计时器timer_，私有方法里添加time_callback回调函数声明。

const rclcpp::TimerBase::SharedPtr timer_;
 
 
void time_callback();

3.实现

实现时在PointCloud2FilterTransformNode的构造函数里，添加timer的初始化。

PointCloud2FilterTransformNode::PointCloud2FilterTransformNode(
  const rclcpp::NodeOptions & node_options)
: Node("point_cloud_filter_transform_node", node_options),
  m_angle_filter{
    static_cast<float32_t>(declare_parameter("start_angle").get<float64_t>()),
    static_cast<float32_t>(declare_parameter("end_angle").get<float64_t>())},
  m_distance_filter{
    static_cast<float32_t>(declare_parameter("min_radius").get<float64_t>()),
    static_cast<float32_t>(declare_parameter("max_radius").get<float64_t>())},
  m_input_frame_id{declare_parameter("input_frame_id").get<std::string>()},
  m_output_frame_id{declare_parameter("output_frame_id").get<std::string>()},
  m_init_timeout{std::chrono::milliseconds{declare_parameter("init_timeout_ms").get<int32_t>()}},
  m_sub_ptr{create_subscription<PointCloud2>(
      "points_in", rclcpp::QoS{10},
      std::bind(
        &PointCloud2FilterTransformNode::process_filtered_transformed_message, this, _1))},
  m_timeout{std::chrono::milliseconds{declare_parameter("timeout_ms").get<int32_t>()}},
  m_pub_ptr{create_publisher<PointCloud2>("points_filtered", rclcpp::QoS{10})},
  m_expected_num_publishers{
    static_cast<size_t>(declare_parameter("expected_num_publishers").get<int32_t>())},
  m_expected_num_subscribers{
    static_cast<size_t>(declare_parameter("expected_num_subscribers").get<int32_t>())},
  m_pcl_size{static_cast<size_t>(declare_parameter("pcl_size").get<int32_t>())},
  timer_{create_wall_timer(200ms, std::bind(&PointCloud2FilterTransformNode::time_callback,this))}

把订阅和发布的初始化放到时间回调函数time_callback中

void PointCloud2FilterTransformNode::time_callback()
{
  m_pub_ptr=this->create_publisher<PointCloud2>("points_filtered", rclcpp::QoS{10});
  m_sub_ptr=this->create_subscription<PointCloud2>(
      "points_in", rclcpp::QoS{10},
      std::bind(
        &PointCloud2FilterTransformNode::process_filtered_transformed_message, this, _1));
}

由此便实现了将节点改为以时间周期驱动的节点。