The ocean reflects the laser beams

[osrf]

The ocean reflects the laser beams

[osrf-migration]

Original comment by Carlos Agüero (Bitbucket: caguero, GitHub: caguero).

---

• changed title from "The ocean reflects the laser beams." to "The ocean reflects the laser beams"

[osrf-migration]

Original comment by Carlos Agüero (Bitbucket: caguero, GitHub: caguero).

---

• changed milestone from "0.3" to "0.5"

[osrf-migration]

Original comment by Jonathan Wheare (Bitbucket: Jonathan Wheare).

---

Hi Carlos,

I have been looking at this and the linked issue. If you want an example of the behavior of a velodyne on the water, the lidar data in the following image was accumulated from several scans of an artificial lake. The sensor was an HDL-32 mounted on a WAM-V. The top of the image includes trees and a beach, the bottom is the water surface. Note that in this image, the velodyne produces only a very few returns from the water surface.

https://csem.flinders.edu.au/competitions/maritime_robotx/images/gallery/LiDAR%20west%20lakes%20data..jpg

A connected issue may be that the GPU velodyne implementation does not currently support variable reflectivity. A bug has been filed about this on the dataspeed bitbucket.

https://bitbucket.org/DataspeedInc/velodyne_simulator/issues/13/same-intensity-values-populated-if-gpu_ray

If the bug is resolved, it may be possible to create the expected behaviour by setting the oceans laser_retro value to zero.

Regards,
Jonathan.

[osrf-migration]

Original comment by Jonathan Wheare (Bitbucket: Jonathan Wheare).

---

HI Carlos,

After some experimentation, the laser_retro tag is supported by the GPU velodyne_lidar plugin when applied to the visual element of the SDF description. However, all generated lidar returns within the maximum and minimum ranges will be returned even if they have a low intensity value. I have modified a version of the velodyne_simulator plugin that will optionally clip points that have an intensity that lies beneath a specified threshold. If the laser retro value of the ocean is set to a negative value, ocean points can then be removed from the lidar return while still detecting objects that have the default laser_retro value of zero. This version of the plugin can be found here: https://bitbucket.org/JonathanWheare/velodyne_simulator/src/master/

Without clipping. In this image the lidar has returned approximately 14,000 points.

with clipping. In this image only approximately 2500 points have been returned by the lidar.

(Note: buoy laser_retro values have been set to 100 to enable them to be clearly distinguished from the ocean return.)

Regards,
Jonathan.

[osrf-migration]

Original comment by Brian Bingham (Bitbucket: brian_bingham).

---

Thank you @JonathanWheare. This seems very promising. I'll let Carlos comment on the details. Do we think that the velodyne_simulator folks will entertain a PR to implement this feature? From your fork it seems like a low-impact addition.

[osrf-migration]

Original comment by Jonathan Wheare (Bitbucket: Jonathan Wheare).

---

Hi Brian,

I think it is a sensible enhancement to the simulator package.

Regards,
Jonathan.

[osrf-migration]

Original comment by Carlos Agüero (Bitbucket: caguero, GitHub: caguero).

---

Hi Jonathan, this solution looks very promising. One more question: Aren't the beams hitting the ocean floor? It's not clear from your picture but I'd expect to happen.

[osrf-migration]

Original comment by Carlos Agüero (Bitbucket: caguero, GitHub: caguero).

---

In any case, I suggest to go ahead and submit your pull request. Looks like a good contribution to velodyne_simulator. I'll share my comments in your pull request. Feel free to tag me.

[osrf-migration]

Original comment by Jonathan Wheare (Bitbucket: Jonathan Wheare).

---

Hi Carlos,

I cannot tell you exactly how far through the water surface that a lidar will give a return, but I can state that my experience has been that such returns have been very rare. The west lakes image I linked to was an accumulation of several passes of the lidar, and beyond the vehicle, shore and test docking area (shown here: https://csem.flinders.edu.au/competitions/maritime_robotx/images/gallery/west_lakes_docking.jpg) only a small scattering of returns can be seen. I interpret these returns as coming from the water surface. There may be places where the water is shallow enough for a return to come from the floor, but in the RobotX course area, I expect such returns would be minimal.

Thanks for your support, I will put together a pull request for velodyne_simulator.

Regards,
Jonathan.

[osrf-migration]

Original comment by Carlos Agüero (Bitbucket: caguero, GitHub: caguero).

---

Thanks Jonathan, looking forward to see the pull request!

[osrf-migration]

Original comment by Carlos Agüero (Bitbucket: caguero, GitHub: caguero).

---

• changed milestone from "0.5" to "1.1"

[osrf-migration]

Original comment by Jonathan Wheare (Bitbucket: Jonathan Wheare).

---

Hi Carlos,

I just sent in pull request that should enable filtering of the returns. Note that by default, other objects in the world will still give zero intensity returns.

Regards,
Jonathan Wheare.

[osrf-migration]

Original comment by Carlos Agüero (Bitbucket: caguero, GitHub: caguero).

---

• changed state from "new" to "resolved"

Solved with https://bitbucket.org/DataspeedInc/velodyne_simulator/pull-requests/7/master/diff and pull request #79. Thanks @JonathanWheare !