• State TI Thinks Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 34 subscribers
  • Views 234 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.

IWR6843ISK: Interframe jumps: Fine motion detection

Girish Vaidya
Prodigy 125 points

Part Number: IWR6843ISK
Other Parts Discussed in Thread: IWR6843

Hi,

  The document 'Detection Layer Parameter Tuning Guide for the 3D People Tracking Demo' describes the process of fine motion detection. The procedure combines k chirps across M frames to create a multi-frame radar cube. This is clear. 

 

I have two queries:

1) How is the doppler computation done with multi-frame radar cube? Since only k chirps from each frame are used, there will be abrupt phase jump after every k chirps. How is this compensated while computing doppler fft?

2) In addition to phase jump due to limited number of chirps, there will be phase jump due to interframe delays. In other words, even if we assume that a multiframe radar cube is created by combining all chirps within a frame, there will be phase jump at frame boundaries. How is this compensated?

Appreciate your help to understand the implementation.

 

Best Regards,

Girish

 

  • 0
    TI__Mastermind 29492 points

    We will get back to you within 24 hours

    Best Regards,

    Pedrhom

  • 0 in reply to Pedrhom Nafisi
    Prodigy 125 points

    Hi Pedrhom,

      Curious if you have any inputs on this.

    Best Regards,

    Girish

  • 0 in reply to Girish Vaidya
    Prodigy 125 points

    Hi Pedrhom,

      Any inputs regarding this.

    Regards,

    Girish

  • 0 in reply to Girish Vaidya
    TI__Expert 5600 points

    Hi,

    the system relies on a combination of techniques to mitigate these phase jump effects. Primarily, the Capon beamforming process itself, used for Doppler estimation, helps to reduce the impact of phase discontinuities. Additionally, our filtering mechanism, which separates dynamic and static points, requires a degree of phase correction to function effectively.

    We're investigating the specific algorithms used in more detail and will follow up.

  • 0 in reply to Sharan S S
    Prodigy 125 points

    Hi Sharan,

       This is not clear to me. As far as I understand, the capon beamforming is for angle estimation. How does this help to reduce the phase discontinuities. Could you please elaborate.

    Do the static points also require phase correction? Or is it implemented only for the dynamic processing chain?

    Looking forward to a detailed response. Could you also point me to the part of the code that performs the phase correction.

    Best Regards,

    Girish

  • 0 in reply to Girish Vaidya
    TI__Expert 3710 points

    Hi Girish,

    Sorry for the delay in response.

    To get more clarity on the doppler computation please also refer to the document present at location:
    radar_toolbox_4_00_00_05\source\ti\examples\Industrial_and_Personal_Electronics\People_Tracking\docs\IWR6843_People_Tracking_PDFs\3D_people_tracking_demo_implementation_guide.pdf

    Here, Section 6.1.2.2.3 (Radial Velocity Estimation) explains briefly how the low doppler points from fine motion mode are forced to be fully static.

    Girish Vaidya said:
    1) How is the doppler computation done with multi-frame radar cube? Since only k chirps from each frame are used, there will be abrupt phase jump after every k chirps. How is this compensated while computing doppler fft?

    In general, there are 2 ways in which we can do minor motion across frames:

    a. Take all the chirps across frames: Here, true doppler computation cannot be done as there is an abrupt phase jump across chirps. But for minor motion, we want to measure minute motion like typing, breathing, or similar actions, which in most use cases, do not require doppler information. Hence doppler can be forced to be zero.

    b. Just take the first burst per frame for minor motion: Here, the chirps will be equidistant across frames and hence, there will be no abrupt jump of phase across frames. Here, doppler can be calculated normally.

    I am presently unsure of which implementation is done in 3D People tracking demo for IWR6843. But you can have a look at the 3D_people_tracking_demo_implementation_guide.pdf to find that.

    (I suspect approach (a) may be used since Section 6.1.2.2.3 mentioned forcing the doppler of fine motion points to be zero)

    Girish Vaidya said:
    2) In addition to phase jump due to limited number of chirps, there will be phase jump due to interframe delays. In other words, even if we assume that a multiframe radar cube is created by combining all chirps within a frame, there will be phase jump at frame boundaries. How is this compensated?


    If we use approach (a) then no compensation is needed here.

    If approach (b) is used, a technique called 'Doppler Phase Compensation' is used to compensate for this. You can again check in the implementation guide if this feature is used.

    Hope this resolves your query.

    Thanks and Regards,
    Saransh Gautam