How to get local position and orientation through IMU by MAVROS

[weiweikong]

Hi there, I am currently working on project where we are implementing a Pixhawk Autopilot and ROS for SLAM with Hokuyo UTM-30LX laser and monocular/stereo camera. Now we did not use external IMU. So the setup of our system are

• Ubutnu 14.04 64bit
• ROS indigo
• Pixhawk Autopilot

I have already use MAVROS receive the data from IMU. After running rosrun mavros mavros_node, there are /mavros/imu/data and /mavros/imu/data_raw on the topic list. But I am not so familiar with Pixhawk and this research filed, so here are some questions:

• 1.The difference between /mavros/imu/data and /mavros/imu/data_raw Regard to mavros/mavros/src/plugins/imu_pub.cpp files,
  • What is the difference between void handle_attitude and void handle_attitude_quaternion ? Which kind of IMU is needed and how could we connected it with Pixhawk?
  • What is the difference between void handle_raw_imu and void handle_scaled_imu?
• 2.From Quaternion to Euler Angle ? I tried to get the Euler Angle (Roll, Pitch , Yaw) using getRPY through Quaternion data from IMU, but the result seems wrong as I rotated Pixhawk. Is that caused by Gamble Lock when the rotation more than 90 degree? And which function should be adpated to access the Euler Angle.
• 3.Electromagnetic Disturbances? Also, I tried to test Pixhawk with PTU as shown below for calbrating Laser and Camera System. The Quaternion data were drifting during the rotation of PTU. Is that the electromagnetic disturbances? How could I access the data directly from IMU without the fusion of compass?
• 4.Access the local position and orientation As mentioned in the paper PIXHAWK: A Micro Aerial Vehicle Design for Autonomous Flight using Onboard Computer Vision, both IMU and vision could calculate the attitude, but how could the system get the local position? Is that possible to get the local position and orientation directly from Pixhawk only with the provided IMU by calculating the accleration and rotation ? Or the only way to get the local position during the flying is depond on Vision System?

Does anyone have any ideas or suggestions?

Thanks.

Best, Weiwei

[vooon]

What pixhawk firmware do you use?

1. Handles different set of messages and consolidate to standartized imu/data and imu/data_raw topics. Where:

   • imu/data - oprientation computed by FCU (converted from Euler for ATTITUDE or Q from ATTITUDE_QUATERNION) plus accel data from last sensor raw data.
   • imu/data_raw - only data from raw messages.

   Note that ROS use ENU frame, while FCU - NED.

2. Did you calibrate IMU?
3. imu/mag.
4. Maybe local_position.cpp?

[weiweikong]

Hi Vooon, thanks for your reply. The Firmware is the stable version, px4fmu-v2_default.px4 for PX4FMU v2.x (in Pixhawk)

• 1.More details,
  • 1.1 Could I get the Euler Angle directly through MAVROS? I checked the code here, it might be no.
  • 1.2 In imu_pub.cpp, what is the difference between void handle_attitude and handle_attitude_quaternion. Is that depend on the hardware of Pixhawk or IMU?
  • 1.3 Regard to tf::Quaternion orientation = tf::createQuaternionFromRPY(att.roll, -att.pitch, -att.yaw);, it seems the output of Quaternion from imu/data is under ENU frame ?
  • 1.4 Also, is the Quaternion fusioned with compass?
• 2.I have already calibrate IMU by Mission Planner.
• 3.Solved.
• 4.There is no reply after typing rostopic echo /mavros/position/local

[vooon]

1. ATTITUDE give RPY angels, and then translated to Quaternion, is body frame. See #49. ATTITUDE_QUATERNION only rotates axis to match ROS ones. Some FCU's provide both of this messages, some not. Also all fusion done in FCU, so compass processing must be in firmware.

[mhkabir]

@weiweikong Any updates on this?

[mhkabir]

Please use the PX4users mailing list or Gitter if you have further questions : https://gitter.im/mavlink/mavros https://gitter.im/PX4/Firmware

[zer1988]

Hi vooon can you guide me how I can convert data of orientation of mavros/imu/data from quaternion to spatial rotaion?

[vooon]

Use tf python lib. Or use Eigen.