• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 3 answers
  • Subscribers 32 subscribers
  • Views 199 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

IWRL6432: question about target z coordinate

? ??
Prodigy 20 points
Part Number: IWRL6432

Tool/software:

Hi Ti expert,

I am using motion_and_presence_detection demo in MMWAVE_L_SDK_05_04_00_01 with GTRACK_3D defined and rebuild image.

When detect an adult with about 170cm height, the posX, posY value in trackerProc_Target is correct, but all the posZ values are small, such as 0.071038.

It's not the height of the human target?

My cfg param:

sensorPosition 0 0 1.9 0 0 

Tracker data exmaple:
time, tid, posX, posY, posZ, velX, velY, velZ, accX, accY, accZ
15:13:40,tid 1, 2.439973, 3.257148, 0.071038, -0.201151, -0.268519, -0.005856, 0.000000, 0.000000, 0.000000
15:13:40,tid 1, 2.419858, 3.230296, 0.070453, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000, 0.000000
15:13:40,tid 1, 2.409565, 3.213783, 0.068255, -0.106357, -0.169849, -0.022166, -0.068619, -0.094470, -0.003819
15:13:40,tid 1, 2.394245, 3.187147, 0.060574, -0.139289, -0.231360, -0.050420, -0.112533, -0.156550, -0.010281
15:13:41,tid 1, 2.389378, 3.141573, 0.050747, -0.122868, -0.330514, -0.068077, -0.142384, -0.232732, -0.015953
15:13:41,tid 1, 2.280830, 3.154223, 0.033395, -0.411359, -0.246547, -0.098230, -0.321213, -0.283924, -0.02793
15:13:41,tid 1, 2.282752, 3.075804, 0.019787, -0.347174, -0.401404, -0.109375, -0.293416, -0.372096, -0.03218
15:13:41,tid 1, 2.246567, 3.033804, 0.008689, -0.188258, -0.219307, -0.056297, 0.000000, 0.000000, 0.000000
15:13:41,tid 1, 2.411302, 2.840402, 0.002690, 0.117605, -0.570647, -0.056762, 0.122749, -0.311548, 0.000177
15:13:41,tid 1, 2.292085, 2.847456, -0.001312, -0.088003, -0.482859, -0.054381 -0.010742, -0.216143, 0.00085
15:13:41,tid 1, 2.326970, 2.757880, -0.004731, -0.009800, -0.566832, -0.051132 0.059893, -0.254722, 0.002914
15:13:41,tid 1, 2.374248, 2.665052, -0.005859, 0.098298, -0.632387, -0.044820, 0.175566, -0.259745, 0.007021
15:13:41,tid 1, 2.384955, 2.600514, -0.010306, 0.057927, -0.329181, -0.022059, 0.000000, 0.000000, 0.000000

  • 0
    TI__Expert 8990 points

    Hi

    Thanks for your query. Please allow us a couple of days to respond.

    Regards

  • 0 in reply to Abhishek Karkisaval
    TI__Genius 16601 points

    Hello, 

    ? ?? said:
    It's not the height of the human target?

    No, the posZ value would not correspond to the target height but rather the z coordinate for the calculated track centroid. However, it still seems strange that the centroid height would be so close to 0. I suspect the configuration may have some issue. 

    Which configuration are you using for the motion_and_presence_detection demo?

    Due to the antenna design for the IWRL6432BOOST EVM, the angular resolution in the elevation direction is not as good as in the azimuth direction. For this reason, most of the default configurations do not enable processing in the elevation direction. 

    To enable this processing and get z values for detected points, you can increase elevation FFT size in the configuration file. For example, using the configuration file motionDetect.cfg ({MMWAVE_SDK5_INSTALL_DIR}\examples\mmw_demo\motion_and_presence_detection\profiles\xwrL64xx-evm), increase elevation FFT size from 2 to 4:

    sigProcChainCfg 32 4 1 0 4 4 0 15
     
    Best Regards,
    Josh
  • 0 in reply to Josh Dye
    Prodigy 20 points

    Hi

    Thank you for reply.

    After changing sigProcChainCfg elevation FFT size from 2 to 4,  z coordinate seems correct.

    For example z coordinate for target 0 is 0.782266,  z coordinate for target 2 is 0.020543

    what's the meaning of the positive and negative values of z coordinate?


    For the "sensorPosition 0 0 1.9 0 0 ",   z coordinate 0.782266 means target centroid height = 1.9 - 0.782266  ?


    2024-09-10 16:27:32: receive tid 0, x 1.112907, y 0.724657, z 0.782266
    2024-09-10 16:27:32: receive tid 2, x -0.834178, y 1.063950, z 0.020543
    2024-09-10 16:27:32: receive tid 0, x 1.112907, y 0.724657, z 0.782266
    2024-09-10 16:27:32: receive tid 2, x -0.794445, y 1.044500, z 0.017359
    2024-09-10 16:27:32: receive tid 2, x -0.737311, y 0.998074, z 0.011536

  • 0 in reply to ? ??
    TI__Genius 16601 points

    Hi, 

    No problem. The target centroid actual z location would be: z coordinate + sensorHeight.

    Best Regards,

    Josh

  • 0 in reply to Josh Dye
    Prodigy 20 points

    Hi

       If target centroid actual z location is coordinate + sensor, then my result height = 1.9 + 0.782266,

    z coordinate 0.782266 is simply not correct.

    But when changing FFT size larger than 4, there is no tracker data output in motion_and_presence_detection demo.

    sigProcChainCfg 32 4 1 0 4 4 0 15

    Do you have any idea about this issue?

  • 0 in reply to ? ??
    TI__Genius 16601 points

    Hello, 

    Please check the tuning guide at {MMWAVE_SDK5_INSTALL}/docs/Motion_Presence_Detection_Demo_Tuning_Guide.pdf. Sections 2.4.3 and 3.2.1 contain relevant information for angle processing and angle FFT size. With a small elevation FFT size your inter-bin resolution is not great. Increasing the elevation FFT size will improve this; however, when you try a value larger than 4 you may be facing a memory constraint issue. Do you see any errors in this case or just that there is no data output?

    Best Regards,

    Josh

  • 0 in reply to Josh Dye
    Prodigy 20 points

    Hi,

    I didn't see any errors while changing elevationFftSize to 8,  just there is no data output.

    From MMWAVE_L_SDK_05_04_00_01\tools\visualizer log, there is no data frame received.

    FYI, if changing elevationFftSize to 4, there is a data frame log like below can be seen from the visualizer log.

    2024-09-19 11:11:04,474 - frame-parser         - DEBUG : {'version': 84148225, 'totalPacketLen': 512, 'platform': 681010, 'frameNum': 4, 'timeCPUCycles': 0, 'numDetectedObj': 6, 'numTLVs': 5, 'subFrameNum': 4294967295}

  • 0 in reply to ? ??
    TI__Genius 16601 points

    Hello, 

    From the tuning guide I mentioned in my last post:

    "These parameters (azimuthFftSize and elevationFftSize) configure the angle search resolution in both azimuth and elevation domains. A larger value will improve the estimated azimuth/elevation accuracy but also increase the total processing time. The user should make a tradeoff and adjust the angle FFT sizes according to the memory, computation power, and average power consumption limits. Changing this parameter may result in the system running out of memory and MIPS. To fit the required processing time into a single frame period, the user must configure the frame period."

    When running the demo with the Industrial Visualizer instead you can see that the configuration produces a memory related error when elevationFftSize is increased to 8 without changing any other parameters. 

    You can compensate for the increased hwa memory bank usage from increasing the elevationFftSize by decreasing the azimuthFftSize. 

    Best Regards,

    Josh