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.

IWRL6432AOPEVM: fall detection support

Part Number: IWRL6432AOPEVM
Other Parts Discussed in Thread: IWRL6432AOP, IWR6843, IWRL6432

Tool/software:

Hi TI Team,

I am currently working with the IWRL6432AOPEVM and would like to evaluate its fall detection capabilities. Could you please confirm if fall detection is supported on the IWRL6432AOP?

If it is supported, I would appreciate it if you could provide the appropriate firmware and configuration (.cfg) files for this use case. I came across the "Fall Detection Using mmWave" reference, but it appears to be specifically designed for the IWR6843AOPEVM.

Looking forward to your guidance.

Thank you,
Bhavya

  • Hello Bhavya, 

    Thank you for your patience. Please refer to this thread on the same topic.

    https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/1491221/iwrl6432aopevm-human-presence-and-fall-detection 

    Best regards,

    Josh

  • Hello ,

    I have gone through the thread and tried to find the fall detection guide and firmware for IWRL6432AOP in radar_toolbox_3_00_00_05. But there is no such thing I am able to find. Everything was related to IWR6843.

    I have also flashed the motion_and_presence_detection_demo_aop.release.appimage from IWRL6432_People_Tracking example in a hope of fall detection.

    Opened the Industrial Visualizer but dint able to observe anything related to fall. Below is the screenshot of my industrial visualizer.

    Can you please verify at your end and let me know if fall detection is supported on IWRL6432AOP. If yes, kindly provide me the example firmware and cfg which you have verified.

    BR,

    Bhavya

  • Hi Bhavya, 

    My apologies for the confusion, the other thread is misleading and you are correct. Currently, the fall detection feature of the visualizer uses 'track height TLV' data which is output from the xWR6843 people tracking demo but this TLV is not available with xWRL6432. So enabling fall detection in the visualizer with  xWRL6432AOP has no effect. 

    However, in the latest Radar Toolbox release (RTB 3.10.0.5) we have added an updated version of the motion and presence detection demo that differs somewhat from the SDK version. Several improvements have been added including a change to the clustering logic which enables 3D clustering (the changes are documented here), this has allowed us to enable a crude fall detection functionality that uses the presence detection/zone logic instead of the tracker. A fall detection cfg file is added at {RADAR_TOOLBOX_3_10_0_5}\source\ti\examples\Industrial_and_Personal_Electronics\Motion_and_Presence_Detection\Motion_and_Presence_Detection\chirp_configs\height_based_detection\xwrl6432aop\fall_detection.cfg. 

    Best regards,

    Josh

  • Hi ,

    Thank you for the response. I have tried fall detection with RADAR_TOOLBOX_3_10_0_5

    My test setup is as per below.

    Firmware: radar_toolbox_3_10_00_05\source\ti\examples\Industrial_and_Personal_Electronics\Motion_and_Presence_Detection\Motion_and_Presence_Detection\prebuilt_binaries\motion_and_presence_3d\motion_and_presence_detection_demo_aop.Release.appimage

    Config file: radar_toolbox_3_10_00_05\source\ti\examples\Industrial_and_Personal_Electronics\Motion_and_Presence_Detection\Motion_and_Presence_Detection\chirp_configs\height_based_detection\xwrl6432aop\fall_detection.cfg

    Visualizer: radar_toolbox_3_10_00_05\tools\visualizers\Applications_Visualizer\Industrial_Visualizer

    Observation:

    Presence and motion I am able to detect. But regarding fall there is no information/indication in the Visualizer tool. I have also enabled the fall detection in the tool. I have tried to test the application by actually falling Slight smile but not able to find any indication in the visualizer tool. Am I missing something?

    I have gone through the "Motion and Presence Detection Users Guide" but there is no information to test fall detection.

    Can you please check and guide me to test fall detection with IWRL6432AOPEVM.

    Moreover which TLV data is used to identify the fall detection as finally I am going to use the sensor with host MCU which will take further decision based on the fall detection?

    BR

    Bhavya

  • Bhavya, 

    I apologize for the misunderstanding as I did not explain fully in my last post. There is no explicit fall detection algorithm running or TLVs related to fall detection. This configuration essentially creates 2 zones. The zone closer to the ground should only be occupied when a person is laying on the ground. So the functionality is more of a 'has fallen' detection as it does not detect the act of someone falling.

    Best regards,

    Josh

  • Hi ,

    Ok understood. I have this demo using fall_detection.cfg but it seems like the zone closer to the ground is not able to detect the occupancy when I was laying down on the ground, Am I missing something to test this demo?

    More over using this scenario it will also give the false signal if the person is actually sleeping on the ground, isn't it?

    Is there any plan in near future to enable support for fall detection in iWRL6432AOP?

    BR,

    Bhavya

  • Hi Bhavya, 

    I have this demo using fall_detection.cfg but it seems like the zone closer to the ground is not able to detect the occupancy when I was laying down on the ground, Am I missing something to test this demo?

    Please confirm the values for sensorPosition in your configuration are correct for your mounting height and angle.

    More over using this scenario it will also give the false signal if the person is actually sleeping on the ground, isn't it?

    Yes, I'm not certain if this specific configuration was tested with sleeping people but that is most likely correct.

    Is there any plan in near future to enable support for fall detection in iWRL6432AOP?

    I don't believe there is a plan to explicitly support fall detection for IWRL6432AOP but we are investigating ways to improve the point cloud which may allow for a more conventional fall detection algorithm to be more successful (such as what is done in IWR6843). For example, one possibility might be to integrate the CAPON beamforming (MVDR) algorithm and tracker (gTrack) algorithm. These algorithms are both available currently but in separate examples (xWRL6432 Life Presence Detection Capon 2D) for the xWRL6432 so some development work would be required to combine them. This is an option we are looking into but I can't say for certain if or when any example might be released.

    Best regards,

    Josh

  • Hi ,

    Please confirm the values for sensorPosition in your configuration are correct for your mounting height and angle.

    I have placed my sensor at 1.3m height

    sensorPosition 0 0 1.3 0 0

    The complete .cfg file is as below, which is the same as fall_detection.cfg Do you think any other parameter I have to change to detect the occupancy from the zone closer to the ground?

    sensorStop 0
    channelCfg 7 3 0
    chirpComnCfg 8 0 0 256 1 23 0 
    chirpTimingCfg 6 18 0 75 60.5
    frameCfg 8 0 403 1 250 0
    guiMonitor 2 0 0 0 0 1 1 0 0 0 0
    sigProcChainCfg 64 32 2 0 4 4 0 .5 1
    cfarCfg 2 4 3 2 0 9.0 0 0.8 0 1 1 1
    aoaFovCfg -60 60 -40 40
    rangeSelCfg 0.1 10.0
    clutterRemoval 1
    antGeometryCfg 1 1 1 0 0 1 1 3 1 2 0 3 2.5 2.5
    compRangeBiasAndRxChanPhase 0.0 1.00000 0.00000 1.00000 0.00000 1.00000 0.00000 1.00000 0.00000 1.00000 0.00000 1.00000 0.00000
    adcDataSource 0 adc_data_0001_CtestAdc6Ant.bin
    adcLogging 0
    lowPowerCfg 1
    rangeSlope 0 15 40
    factoryCalibCfg 1 0 30 0 0x1ff000
    mpdBoundaryBox 1 -1.48 1.48 0 10 0.25 3
    % mpdBoundaryBox 2 0 1.48 1.95 3.9 0.25 3
    % mpdBoundaryBox 3 -1.48 0 0 1.95 0.25 3
    % mpdBoundaryBox 4 -1.48 0 1.95 3.9 0.25 3
    mpdBoundaryBox 2 -1.48 1.48 0 10 0 0.25
    % mpdBoundaryBox 6 0 1.48 1.95 3.9 0 0.25
    % mpdBoundaryBox 7 -1.48 0 0 1.95 0 0.25
    % mpdBoundaryBox 8 -1.48 0 1.95 3.9 0 0.25
    sensorPosition 0 0 1.3 0 0
    minorStateCfg 8 7 63 8 4 30 8 8
    steerVecCorr 0 0 0
    clusterCfg 1 0.8 2
    baudRate 1250000
    sensorStart 0 0 0 0

    This is an option we are looking into but I can't say for certain if or when any example might be released.

    Okay, thank you for the information. Kindly let me know if any new updates will be available regarding fall detection.

    BR,

    Bhavya

  • Hi Bhavya, 

    Sorry for the delayed response.

    I have placed my sensor at 1.3m height

    sensorPosition 0 0 1.3 0 0

    Thank you, and just to confirm this sensorPosition command indicates that the EVM is mounted perpendicular with the ground (not tilted forward at all). Does that also match your setup?

    I have some questions which may help me provide better suggestions for tuning the configuration. When using the 3D plot in the visualizer, can you see any detected points when laying down or there are very few/no points detected? If you do see some detections but the zone still does not indicate presence is detected by changing color, can you check the height of the detected points relative to the zone's upper and lower boundaries? You can turn rotate the 3D plot to determine if the points are showing up above the zone or under the ground plane.

    Best regards,

    Josh