• State TI Thinks Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 941 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AWR1642BOOST: Point cloud doesn't show points in the middle of the FOV

Jose Castanheira
Prodigy 165 points
Part Number: AWR1642BOOST
Other Parts Discussed in Thread: IWR1642, IWR6843ISK

I tested the people count demo, and I noticed that points related to the reflections of people specifically in front (not on the sides) of the radar don't appear in point cloud. This is more visible the further you are from the RADAR. Furthermore, in the TLV regarding the point cloud there is no information about points located next to the middle axis of the FOV.

Is this supposed to happen or is it a bug?

  • 0
    TI__Mastermind 28055 points
    Hi Jose,

    Please ensure you are using the latest version of the People Counting demo that is available to you.

    I want to confirm your issue: You are not being detected when standing directly in front of the sensor. This bug was noticed and corrected earlier this year. Industrial Toobox 2.3.0 and up will have this bug fixed.

    Please let me know if you are experiencing this bug and are on a newer version of the Industrial toolbox.

    Regards,
    Justin
  • 0 in reply to Justin Curewitz
    Prodigy 165 points
    Hello,

    Thank you for your reply. I am using Industrial Toolbox 2.5.2 and I'm not detected standing directly in front of the sensor.

    Best regards,
    Jose Castanheira
  • 0 in reply to Jose Castanheira
    TI__Mastermind 28055 points

    Hi Jose,

    The demo does not detect targets that are not moving - we use a static clutter removal algorithm to filter out things like tables, chairs, and walls. When you are standing in front of the sensor, are you still? Or do you notice this issue when you cross boresight? If you notice this when you cross boresight, are you walking in a straight line or an arc?

    The GUI has two output windows - "point cloud" and "Gating and Association" - do both of these screens show nothing when you are standing directly in front of the sensor?  Is it possible to provide video of the failure?

    Regards,
    Justin

  • 0 in reply to Justin Curewitz
    Prodigy 165 points
    Hello,

    I am quite aware that the demo does filter static clutter. I notice this issue when I cross boresight, and it happens when I walk in a straight line or in an arc.
    The problem is related to the point cloud; I don't see any points being plotted in the middle of the FOV when there are moving objects in the middle. However, these same objects appear in the "Gating and Association" window, but that is just a consequence of the tracking algorithm used. If the target was detected outside the the "blind zone", then it will still be detected in the "Gating and Association" window when this same object is in the blind zone.

    Best regards,
    José Castanheira
  • 0 in reply to Jose Castanheira
    TI__Mastermind 28055 points
    Hi Jose,

    Currently, there is a blind zone at the 0 degree bin. This is a bug in the detection software. However, from my tests, a person standing at boresight will still generate points to the left and right of the zero degree mark, maintaining detection even if the person stops there. The point cloud is dense enough to generate new tracks as well.

    Regards,
    Justin
  • 0 in reply to Justin Curewitz
    Prodigy 165 points
    Hi Justin,

    You said "a person standing at boresight will still generate points to the left and right of the zero degree mark", but, since we are talking about an angle, the further you are from the sensor, the bigger the blind zone will be. Thus, a person standing far enough of the sensor, at the zero degree bin, will be completely invisible to the sensor.
    Is there a way to fix this? Or will I have to wait for an update on the toolbox that will hopefully fix this bug?

    Best regards,
    José Castanheira
  • 0 in reply to Jose Castanheira
    TI__Mastermind 28055 points
    Hi Jose,

    This issue has been fixed on the IWR6843ISK device, but has not yet been addressed on the IWR1642.

    Regards,
    Justin
  • 0 in reply to Justin Curewitz
    Prodigy 165 points
    Hi Justin,

    I would like to see if this issue is related to the static clutter removal feature, present in this lab. Therefore, I will disable the static clutter removal, and test it on the sensor and see if the point cloud is still "invisible" on the 0 degree range bin . However, looking at the source code, I can't seem to find the right place where I can disable this feature. I would appreciate your insight on this.

    Thanks in advance,
    José Castanheira
  • 0 in reply to Jose Castanheira
    TI__Mastermind 28055 points
    Hi Jose,

    You can find the code that does the static clutter removal here:
    mmwave_industrial_toolbox_2_5_2\labs\lab0011-pplcount\lab0011_pplcount_pjt\radarDemo\modules\DoA\CaponBF\src\RADARDEMO_aoaEstCaponBF_heatmapEst.c

    You can simply change the if at line 118 to if (0) to remove clutter removal.

    This code essentially averages each range bin across chirps, then subtracts the average. What is left is the phase difference, which relates directly to motion.

    Regards,
    Justin