a clear coordinate system definition

[JokerJohn]

Hi, thanks for your work. Can you give a clear coordinate system definition file, similar to the KITTI dataset, which I did not find on the official website?

[hilti-challenge]

Hi, We are currently working on a similar definition.

[hilti-challenge]

Hi, here you can see the diagram:

[JokerJohn]

According to the debugging results of our LIO algorithm in the datasets(exp04/05/06/calib_03), the coordinate system of the lidar and IMU is confirmed as shown in the figure below, in which the coordinate system of the Pandar lidar is different from the REP105 convention. Its xyz correspond to the left, back, and up directions, respectively. It is slightly different from the diagram you provided.

[gogojjh]

An additional comment on the calibration diagram. In this diagram, the X, Y, Z axes of the LiDAR are indicating forward, leftward, and upward. It does not fit the calibration parameters in lidar_calibration.yaml. In this file, the X, Y, Z axes should indicate leftward, backward, and upward respectively.

[mcamurri]

@gogojjh and @JokerJohn you are both correct. I have also updated the corresponding URDF, here: https://github.com/Hilti-Research/hesai_description/commit/64633725c701c2763c5787257f4e32b7bc3067d9 I have updated the diagram, sorry about the inconvenience.

Regarding the conventions, yes the recommendation 105 is to have x-forward, y-left and z-up. However, the raw data are expressed according to the manufacturer's specifications of the sensors (which are unfortunately different), and we preferred to keep them as they are instead of manipulating them by adding extra rotations.

[gogojjh]

Thanks a lot!

[YoshuaNava]

@mcamurri In the commit you linked you don't use the frame name in the Lidar message ("PandarXT-32"), but rather "pandar" ->

Is this on purpose?

[mcamurri]

Sorry because the default value from the driver is pandar, I changed that without realising the frame in the HILTI rosbags is PandarXT-32. I'm going to fix that now.

[YoshuaNava]

@mcamurri I get the following error after trying to launch view_urdf.launch from the packages linked in this repo's README:

[mcamurri]

@YoshuaNava my apologies, can you please try this branch here. If that works, I will merge it.