IWR6843AOP: Issue on TARGET LIST generated based on overhead people counting lab

Hello Kevin,

What do you mean by the Z direction? The orientation of the coordinates can shift depending on the world vs sensor coordinates. I think you mean the elevation direction of the sensor. 

Either way, please see the tracking tuning guide. You may need to reduce the SNR threshold for the tracker or the min points for allocation. Please see the allocationParams and trackingCfg parameters in the following document.

3D_people_counting_tracker_layer_tuning_guide.pdf

Regards,

Jackson

Hi Jackson,

1. The Z axis is the value which had been computed by range and sine of elevation as followings, 

    based on SENSOR view with celling mounted  (not WORLD view here)

  Based on IWR6843 3D(r, az, el) -> (x, y, z)
  r : range
  el: elevation φ     
  az: azimuth   θ  

  z = r * sin(φ)
  x = r * cos(φ) * sin(θ)
  y = r * cos(φ) * cos(θ)




2. the key issue on Target List generated is NOT symmetric performance on both sides of Z axis
   above mentioned Test Case 2 is better than Test Case 1

3. I had tested some parameters on what you mentioned that tuning on allocationParams but not significant improvement

   Any method to tune as symmetric performance on Target List generated?


Regards,
Kevin  
   

Ok thank you for the clarification, I understand now. The matlab visualizer should save a .mat file named fhist_xx.mat. If you look at this it should show you the data of the point cloud. Then you can calculate whether there is a difference in point cloud strength on one side or the other.

One thing to check, is if the issue is related to scene clutter. Have you tried testing in a different environment or rotating the device 90 or 180 degrees? Sometimes there can be some clutter or items causing reflections that will cause these differences in point cloud density or SNR or other metrics. Have you done this?

Regards,

Jackson

Hi Jackson,

I had rotated the device in different degree of 180 degree then the result is still the same as before

not  SYMMMETRIC performance along with Z-axis.

(1) Worse Case:

      If a man stands at the location of  "Z = +3.0 Meter" then lots of point cloud can be seen but the TARGET LIST

      could not be generated until walking approach to "Z = +1.5 Meter" I can see one TRAGET be genrated

(2) Better Case:

     If a man stands at the location of  "Z = -3.0 Meter" then lots of point cloud can be seen and the TARGET LIST

     could be generated I can see one TRAGET be genrated

Why does the device have the unSymmetric beharivor?      

Regards,

Kevin

Hi Kevin,

This is difficult to say. Do you have the saved fHist file with the pointcloud and tracking data? If you can upload that here I can take a look in more detail.

Regards,

Jackson

Hi Jackson,

The logged file(fHistRT_0001.mat) and snapshots as followings,

(1) upload file fHistRT_0001.mat

     Please let me know whether upload success ? (by using drag and drop method,  drag file over to the "Description" box on E2E)

(2) walikng direction from +Z to -Z axis (worse case, Target List Generated started from +1 M)

(3) walikng direction from -Z to +Z axis (better case, Target List Generated started from -2 M)

Regards,

Kevin

Hi Kevin,

Looks like it didn't upload properly. Try using the insert>file. I can't tell much from the images.

Thanks,

Jackson

Hi Jackson,

reSend attached file,

Regards,

Kevin

Hi Kevin,

Unfortunately, I can't access dropbox or other filesharing applications. The best way is to attach the file here. Did you try using the insert>file option? That has worked for me before.

Regards,
Jackson

Hi Jackson,

Try to using insert>file option

It is not allowed to be inserted see following,

Regards,

Kevin

Hi Kevin,

I think instead of copying in the file location, if you click the upload button it will let you choose. If not, please try adding it to a zip file and uploading here. Not sure why it isn't working for the other methods.

Regards,

Jackson

Hi Jackson,

Do you have another way to load file of 'fHistRT_0001.mat'?

Reagrds,

Keivn

Did you try putting it in a zip file?

Also please make sure you are trying this on Google Chrome browser.

Also, it seems you might be trying to upload the link from dropbox. Please make sure it is a file located on your local machine.

Hi Jackson,

It seems accept ZIP file format, see following,

fHistRT_0001.zip

Regards,

Kevin