• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 30 subscribers
  • Views 321 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.

Original question:

IWR1443BOOST: Cutting the static and non existing object from point cloud represnetation

IWR1443BOOST: Behind glass detection + Permanent detection

Abdellah Bouchama
Prodigy 160 points
Part Number: IWR1443BOOST
Other Parts Discussed in Thread: IWR1443, IWR6843

Hello TI team,

I'm having three questions around same subject, I've already asked some related question in this thread.

I'm always having the same problem of detecting the glass wall in my testing environment.

I've used two type of installation (2 meters elevation 15° tilt and 2 meters elevation and 25° tilt)

When using the 15° i see on the Rviz screen that i detect the glasses but when using the 25° one it disappears.

My questions are :

  • Why does the tilt affect the detection of glass walls?
  • Is there any configuration may help me to avoid glass wall detection? (I've already seen this and it's not compatible with my device).
  • I noticed that bodies behind the glass wall are not detected, how can i activate detection behind glass walls?
  • The bodies are detected only on movement (if it static it's not detected). could you please explain the parameter that i should tune to make the tracking of moving objects active (knowing that i use the OOB image on the IWR1443BOOST also I'm using the occupancy config file for same device mentioned in the previous thread mentioned above).

I'm very near to my final solution, having answer for these question may really help me finalizing my configuration.

Please note that my application is for human body tracking (for pose estimation).

Have a good day.

Abdellah.

  • +1
    TI__Mastermind 22577 points

    Hello Abdellah,

    • Why does the tilt affect the detection of glass walls?

    This is due to the elevation/azimuth of a sensor with reference to the object detected mixed with the difference in reflectivity. The reflection of an RX wave will be different when hitting an object straight on/perpendicularly versus hitting it at an angle.

    • Is there any configuration may help me to avoid glass wall detection? (I've already seen this and it's not compatible with my device).
    • I noticed that bodies behind the glass wall are not detected, how can i activate detection behind glass walls?

    For both these questions: you are slightly limited with an IWR1443, but your best bet is a combination of increasing TX power and decreasing the CFAR threshold. Do not that when you decrease the CFAR threshold, you will be seeing more points so although you will more accurately track bodies behind a glass wall, it is possible that you will detect the glass door itself even more. That is where changing the TX power comes in to compensate for this decreased threshold. It is a very fine balancing act and the right values are different from case to case.

    • The bodies are detected only on movement (if it static it's not detected). could you please explain the parameter that i should tune to make the tracking of moving objects active (knowing that i use the OOB image on the IWR1443BOOST also I'm using the occupancy config file for same device mentioned in the previous thread mentioned above).

    One technique you can try on a IWR1443 is disabling clutterRemoval. When you do this you will probably see an immediate increase in the amount of points you are seeing and will likely need to decrease TX power and increase CFAR threshold to compensate for the static points you are seeing.

    Best Regards,

    Pedrhom Nafisi

  • 0 in reply to Pedrhom Nafisi
    Prodigy 160 points

    Hello Pedrhom,

    Thank you for answers,

    I would like to confirm that the clutterRemoval is already disabled, and the CFAR threshold is in the optimum value (i will give it a new tuning try with the TX Power).

    However, could you please explain to me how to set the TX Power? (i'm beginner TI radar user)

    when you say : ''"you are slightly limited with an IWR1443"" , are you confirming that a device like IWR6843 is better for this kind of use cases (if any other please let me know)?

    Another related question: do not you think that using virtual antennas will help to improve detection performances? (i found this but i do not understand the answer)

    Have a good day.

    Abdellah.

  • 0 in reply to Abdellah Bouchama
    TI__Mastermind 22577 points

    Hello Abdellah,

    TX Power can be adjusted via profileCfg with the txOutPower parameter. Note that this parameter is TX Power BACKOFF which means the higher the value the lower your TX Power, and with the value set to 0 you have your highest TX Power. So you can increase this value while lowering CFAR threshold to test performance. Also know that TX Power is in dBm and we are giving the sensor a decimal value so for example 657930 is a 10dBm TX Power backoff across all three antennas. You can use this thread to understand how to set TX Power better and might help you understand a little more compared to the link you showed: https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/1077964/awr1843-how-can-i-backoff-tx-power-reduce-rx-gain

    IWR6843 is all around better and still actively supported and improved on by TI, so yes I confirm that our IWR6843 is the best device we offer currently for your use-case.

    Best Regards,

    Pedrhom Nafisi