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.

[FAQ] AWR1642: Implementing people count on AWR1642BOOST-ODS

Part Number: AWR1642
Other Parts Discussed in Thread: IWR6843, IWR1642BOOST


I'm trying to implement people count demo on AWR1642BOOST-ODS. I made following modification on chirp config :

  • "channelCfg 3 3 0" to  activate RX1, RX2, TX1 and TX2, this makes a virtual antenna array of size 4 which is allowed by the demo
  • "SceneryParam 0 3 0.05 4" I put the board on the corner of my room

  • "AllocationParam 10 0.01 15 1 2" due to lower antenna gain on AWR1642BOOST-ODS I decreased minimum total SNR to 10 and minimum number of points to 15

Then I ran the demo , it gave me a rich point cloud on visualizer when I walked in front of it but it didn't assign any track. Please tell me how to configure chirp configuration to make it work on AWR1642BOOST-ODS.



  • Hi Hossein,
    Is there any specific reason you are moving to AWR1642-ODS for people count? This demo has been done and tuned for IWR1642-BOOST and IWR6843 EVM.
    So I would suggest to use one of above EVM.

  • Hi Jitendra

    I bought AWR1642BOOST-ODS because I want to implement people counting demo in 3D space, that time IWR6843 wasn't exist and only AWR1642BOOST-ODS has the antenna pattern that allows 3D processing. My current goal is to implement 2D people counting demo on AWR1642BOOST-ODS then upgrade it. Beside the only difference between AWR1642BOOST-ODS and IWR1642BOOST is the antenna pattern so technically it should be possible to implement people counting demo on both boards.

  • Hi Hossein,

    The People Counting lab is developed for a radar system that has a virtual antenna pattern of 8 Virutal Rx in a row. The Signal chain works as follows:

    1D Range FFT - 2D Capon Beamforming for Range/Angle Heatmap - CFAR on Range Angle/Heatmap - Doppler FFT for velocity of each detected point

    As you can see, both Angle of Arrival (AoA) and Detection in this chain rely on having the assumed Virtual Rx pattern. When running this demo on 1642ODS, you have a different Virtual Rx pattern - as a result, the heatmap and detected points are incorrect.  You will still have detections, but these will not correlate welll with real world locations of objects, resulting in the tracker failing.

    To make the demo work in 3D, you will need to do the following:

    1. Get a 3D point cloud built for the ODS antenna pattern.
    2. Interface point cloud with MSS Tracker Application
    3. Compile the Tracker to work in 3D

    For step 1, you can use the OOB demo for the ODS.

    For step 2, please see:

    1. : <Industrial_Toolbox_3.0.0>/labs/lab0015_pplcount_68xx/common/mmw_output.h, which describes the expected format of the point cloud output
    2. : <Industrial_Toolbox_3.0.0>/labs/lab0015_pplcount_68xx/common/mmw_messages.h and  <Industrial_Toolbox_3.0.0>/labs/lab0015_pplcount_68xx/common/mmw_config.h which show other shared info between the DSS and MSS.
    3. <Industrial_Toolbox_3.0.0>/labs/lab0015_pplcount_68xx/mss/task_mbox.c which is the MSS task that receives data from the DSS

    For step 3, please:

    1. <Industrial_Toolbox_3.0.0>/labs/lab0015_pplcount_68xx/mss/gtrack and copy this into <mmwave_sdk>\packages\ti\alg
    2. Compile the tracker:
      1. In <mmwave_sdk>\packages\ti\alg\gtrack, open the command terminal
      2. Run Setenv.bat
      3. Run "gmake gtracklib3d
    3. Ensure the MSS links the new folder.



  • Hi Justin

    Thanks for your detailed response. I will implement the instruction and post the result here.

    Best regards