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.

IWR6843AOPEVM: default processing and orientation of radar sensor data

Part Number: IWR6843AOPEVM
Other Parts Discussed in Thread: MMWAVEICBOOST

Hello

We are trying to analyze data captured from our IWR6843AOPEVM (with MMWAVEICBOOST). We have optimized a profile using the mmWave Demo Visualizer. and are capturing data using matlab. Now we are struggeling to map the point cloud to the room coordinates. Unfortunately we could not find relevant information on the axes of the received radar data cube.

Can you point us to the documentaion or provide us information on the exact orientation and positioning? Is there some pre-processing being done regarding axes on the sensor already? Or are there some assumptions we should know (default offset, tilt, etc.)? (e.g. in https://e2e.ti.com/support/sensors/f/1023/t/899463

  • Dear QumJo:

    Are you getting your data to MATLAB via USB cable in a similar fashion to the OOB setup?

    Best regards,

    Connor Desmond

  • Hi QumJo

    I don't know exactly what you mean with OOB. But I assume the answer is yes.

    Thanks

  • QumJo:

    OOB-> Out of Box Demo. Can you explain why you are using radar cube data?

    Best regards,

    Connor Desmond

  • Hi Connor

    Pardon, Radar data cube is the wrong term. We are using Matlab and the pointcloud data provided.

    The script we are using from the demo (from TI, called „area_scanner_visualizer“) reads as follows:

    if(SETUP_VIA_GUI)

        hApp = setup_as(); %use _exported for 2017a support; setup_as() for 2019+

    else

        % Manual/programmatic entry

        REAL_TIME_MODE = 1; %1 for from device 0 for playback from dat file

        ENABLE_RECORD = 0;

        datFile.path = '';

        datFile.name = 'areascanner_demo_uart_stream.txt';

        cfgFile.path = '';

        cfgFile.name = 'profile.cfg';

        logFile.path = '';

        logFile.name = 'areascanner_demo_uart_stream.txt';

        offset.height = 1.6;

        offset.el = 0;

        comPort.cfg = 5;

        comPort.data = 4;

        comPort.status = 0;

        zones.enable = 0;

        zones.criticalStart = 0;

        zones.criticalEnd = 1;

        zones.warnStart = 1;

        zones.warnEnd = 4;

        zones.projTime = 2;

    end

    offset.az = 0; % TODO need to add back azimuth offset for user input

    So question is whether there are additional offsets calculated on the dsp or any other assumptions (e.g. a tilted sensor, mounting at a certain height) for this demo.  What are the origin positions for 'offset.height', 'offset.el' and ‚offset.az‘ and what is the sensors assumed orientation?

    Thanks!

  • QumJo:

    Here is a path which describes the coordinate system of the Detected Objects TLV aka Point Cloud.

    <SDK INSTALL PATH>/packages/ti/demo/xwr68xx/mmw/docs/doxygen/html/index.html

    Best regards,

    Connor Desmond

  • Hi Connor

    Thanks for pointing out the location of this documentation. I understand now the coordinate system, but it is still not clear where is the origin in the real world and how should the parameters 'offset.height', 'offset.el' and ‚’offset.az‘ be used.

    Could you please try to answer the following questions or point me to the relevant documentation:

    1. Is the origin (0,0,0) assume to be on the ground/floor of the scene?

    2. Do the angles φ and θ, as shown in the ‘Coordinate Geometry’ documentation, assumed to be zero?

    3. Is 'offset.height' the offset on the z-axis from the origin? In what units?

    4. Is 'offset.el' the offset of φ? In what units?

    5. Is ‘offset.az' the offset of θ? In what units?

    Thanks in advance!

  • QumJo:

    The origin of the scene is the radar. All the units of distance (x,y,z) are in meters. The data from the point cloud are x,y,z spatial measurements relative to the radar. If you want to have another origin, then you will need to apply offsets that are specific to your scene onto the generated point cloud data. If you have x,y,z then you can calculate the spherical coordinates that are referenced in the link that I gave you in the previous reply.

    Best regards,

    Connor Desmond