ROS best practise: tf listener instead of extrinsics params?

[greymfm]

I noticed the extrinsics (for frames base to lidar and base to imu) are in a config file - A more ROS-compatible way would be to directly read them from the ROS transformation tree. I have tested the idea (code below) and it works - maybe I should file a pull request for it and make the idea optional?

Tested with my robot simulator (ROS1 noetic, MID360 simulation):

In 'odom.cc' the existing extrinsics config code for 'extrinsics.baselink2lidar_T' is replaced:

  tf::TransformListener tf_listener;

  // wait for transform: base_link -> imu_frame
  ROS_INFO("DLIO: waiting for transform: base_link -> imu_frame");
  tf::StampedTransform tf_b2i;
  try {
    tf_listener.waitForTransform(this->baselink_frame, this->imu_frame, ros::Time(0), ros::Duration(5.0));
    tf_listener.lookupTransform(this->baselink_frame, this->imu_frame, ros::Time(0), tf_b2i);
  } catch (tf::TransformException &ex) {
    ROS_ERROR("Transform base_link -> imu_frame not found: %s", ex.what());
    ros::shutdown();
    return;
  }

  // base_link -> imu_frame
  this->extrinsics.baselink2imu.t = Eigen::Vector3f(
    tf_b2i.getOrigin().x(),
    tf_b2i.getOrigin().y(),
    tf_b2i.getOrigin().z()
  );

  tf::Matrix3x3 tf_rot_b2i(tf_b2i.getRotation());
  Eigen::Matrix3f R_b2i;
  for (int i = 0; i < 3; ++i)
    for (int j = 0; j < 3; ++j)
      R_b2i(i, j) = tf_rot_b2i[i][j];

  this->extrinsics.baselink2imu.R = R_b2i;
  this->extrinsics.baselink2imu_T = Eigen::Matrix4f::Identity();
  this->extrinsics.baselink2imu_T.block(0, 3, 3, 1) = this->extrinsics.baselink2imu.t;
  this->extrinsics.baselink2imu_T.block(0, 0, 3, 3) = this->extrinsics.baselink2imu.R;

  // wait for transform: base_link -> lidar_frame
  ROS_INFO("DLIO: waiting for transform: base_link -> lidar_frame");
  tf::StampedTransform tf_b2l;
  try {
    tf_listener.waitForTransform(this->baselink_frame, this->lidar_frame, ros::Time(0), ros::Duration(5.0));
    tf_listener.lookupTransform(this->baselink_frame, this->lidar_frame, ros::Time(0), tf_b2l);
  } catch (tf::TransformException &ex) {
    ROS_ERROR("Transform base_link -> lidar_frame not found: %s", ex.what());
    ros::shutdown();
    return;
  }

  // base_link -> lidar_frame
  this->extrinsics.baselink2lidar.t = Eigen::Vector3f(
    tf_b2l.getOrigin().x(),
    tf_b2l.getOrigin().y(),
    tf_b2l.getOrigin().z()
  );

  tf::Matrix3x3 tf_rot_b2l(tf_b2l.getRotation());
  Eigen::Matrix3f R_b2l;
  for (int i = 0; i < 3; ++i)
    for (int j = 0; j < 3; ++j)
      R_b2l(i, j) = tf_rot_b2l[i][j];

  this->extrinsics.baselink2lidar.R = R_b2l;
  this->extrinsics.baselink2lidar_T = Eigen::Matrix4f::Identity();
  this->extrinsics.baselink2lidar_T.block(0, 3, 3, 1) = this->extrinsics.baselink2lidar.t;
  this->extrinsics.baselink2lidar_T.block(0, 0, 3, 3) = this->extrinsics.baselink2lidar.R;

Also, I commented out the transformation publishing (transform: baselink to lidar and transform: baselink to imu) in the DLIO code as my simulation already does this.