Angular velocity is composed differently in the IMU msg versus the Velocity msg

[crazy-harry]

sensor_msgs::msg::Imu imu(const NComRxC *nrx, std_msgs::msg::Header head) {
  ....
  ....
  ....
  // Rotate angular rate data before copying into message ----------------------
  veh_w = tf2::Vector3(nrx->mWx, nrx->mWy, nrx->mWz);
  veh_w *= NAV_CONST::DEG2RADS;
  imu_w = tf2::quatRotate(q_vat, veh_w);
  msg.angular_velocity.x = imu_w.x();
  msg.angular_velocity.y = imu_w.y();
  msg.angular_velocity.z = imu_w.z();
  msg.angular_velocity_covariance[0] = 0.0; // Row major about x, y, z axes
  // ...
  msg.angular_velocity_covariance[8] = 0.0;
  ....
  ....
  ....
  return msg;
}

geometry_msgs::msg::TwistStamped velocity(const NComRxC *nrx,
                                          std_msgs::msg::Header head) {
  auto msg = geometry_msgs::msg::TwistStamped();
  msg.header = std::move(head);

  // Construct vehicle-imu frame transformation --------------------------------
  auto q_vat = getVat(nrx);
  auto r_vat = tf2::Matrix3x3(q_vat);
  auto veh_v = tf2::Vector3(nrx->mIsoVoX, nrx->mIsoVoY, nrx->mIsoVoZ);
  auto veh_w = tf2::Vector3(nrx->mWx, nrx->mWy, nrx->mWz);
  auto imu_w = tf2::Vector3();
  auto imu_v = tf2::Vector3();
  auto q_iso_oxts = tf2::Quaternion();

  q_iso_oxts.setRPY(180.0 * NAV_CONST::DEG2RADS, 0.0, 0.0);
  auto r_iso_oxts = tf2::Matrix3x3(q_iso_oxts);

  imu_v = r_vat * r_iso_oxts * veh_v;
  imu_w = r_vat * veh_w;
  ....
  ....
  ....

  return msg;
}

When parsing and building both the IMU and velocity messages, the vehicle frame angular velocity (veh_w) is initialized the same with the <mWx, mWy, mWz> measurements, but the IMU message has this converted to radians while the velocity veh_w skips this step.

Is there any reason why the resulting imu_w is using radians/sec in the IMU message vs the velocity message using deg/sec? They are both in the INS/IMU frame

[ljones-oxts]

Hi @jake-from-safeai

Thanks for your question. To my knowledge, there is no reason why they would be encoded differently. Indeed the only possible reason why one is encoded in deg/s would be due to our desktop software's preference for deg/s as a user-friendly unit of angular rate.

My understanding is that, as an ecosystem, ROS2 prefers the rad/s unit, am I correct? In which case, my gut instinct tells me the fix would be to encode both in rad/s.

Thanks

Llyr Jones

[crazy-harry]

Given the ROS2 msg output from the driver has both the IMU and Velocity msgs embedded in it, I would think they would atleast in the same units. Could be pretty confusing for someone reading the output of the driver