optimising Euclid depth cam to prioritize close-range objects

[kfind]

I have found that the Euclid's default settings mean that it has trouble seeing small-ish objects (12 x 12 cm, 4cm high) at less than 1m. By trouble, I mean the data is missing/black in the depth image where the object should be. Only when the camera is moved back to about 1 or 1.1m away, does the object finally appear in the depth image data.

Is this something that changing the depth gain or providing an ambient IR flood light might help with? I played with the depth and exposure parameters of the R200 way back, and it did help with similar circumstances, just not sure if the Euclid's camera technology/settings work in the same way, or if someone else has found a good combination of these parameters for their own projects.

[idata]

Hi KFindlater,

Thanks for your interest in the Intel Euclid platform. I think you can try lowering the depth gain (in ROS is r200_lr_gain) to 100, it improves the camera's performance, also try to run the dynamic calibration guide: https://click.intel.com/media/productid2100_10052017/dynamic_calibration_guide.pdf https://click.intel.com/media/productid2100_10052017/dynamic_calibration_guide.pdf. Let us know if it helps. Regards,-Yermi A.

[kfind]

Hello Yermi,

Thanks for that information. Is there a way with the librealsense running on the Euclid with the C++ API (not involving ROS at all) to read and write the gain?

[idata]

Please take a look at the following information that defines general configuration controls in librealsense: https://github.com/IntelRealSense/librealsense/blob/master/include/librealsense/rs.hpp https://github.com/IntelRealSense/librealsense/blob/master/include/librealsense/rs.hpp.

Hope this information helps. Regards,-Yermi A.

[kfind]

So from my testing, I was able to get good quality depth data as close as 38-40cm away, by setting the gain to 100 and by using the 320x240 pixel, 60fps depth stream.

[GSR]

How and where do you change the resolution?

[idata]

To change the resolution you have to change the color_height and color_width parameters in this launch https://github.com/IntelEuclid/realsense_camera/blob/master/launch/lr200m_nodelet_default.launch https://github.com/IntelEuclid/realsense_camera/blob/master/launch/lr200m_nodelet_default.launch. The file can be edited directly or using the Euclid browser interface. The easiest way to do this using the Euclid browser interface is described in the attached document.

[kfind]

Using the software API (in my case, Librealsense in linux) you enable/start the stream either by specifying all of the settings manually, or what I do is choose rs::preset::highest_framerateand it automatically goes down to 320x240 resolution and 60 fps.

for example a line in my C++ code is:

dev.enable_stream(rs::stream::depth, rs::preset::highest_framerate);