• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 561 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.

IWR6843AOPEVM: The chips feeding to doppler estimation module in fine motion detection mode do not have equal interval in time domain

HONGYU YAN
Prodigy 100 points
Part Number: IWR6843AOPEVM

Hi,team

There is a problem confused me when studying the detection leyer of 3D People Counting demo.

In fine motion detection mode,N chirps from M frames are collected,and after static clutter removal,they are used to do fft per range bin per detected point to extract doppler.But as far as I know,the prerequisite for fft to work properly，is the input should have equal interval in time domain.However,in the process described above，the chirps from M frames don't evenly spaced  in time domain。Could you explain about it?

  • 0
    TI__Mastermind 37805 points

    Hi,

    The chirps are approximately evenly spaced in time, if not exactly. Some error is introduced, but it's not debilitating.

    Best,

    Nate

  • 0 in reply to Nathan Block
    Prodigy 100 points

    Hi，Nathan

    Thanks for the reply.

    From the paragraph below,It seems to me the first K chirps has interval of one chirp period,let's say 80us,while the time interval between (K+1)th chirp and Kth chirp is nearly one frame period,for example 50ms.So I don't think those N chirps from M frame are evenly spaced in time.Is there any misunderstanding here?

    " When creating the fine-motion mode radar cube using chirps across multiple frames, the first K chirps
    per frame is utilized (instead of any subsampling) to keep the maximum unambiguous velocity the same
    as the single-frame block.     " from Page 38, Detection layer parameter tuning guide for the 3d people couting demo

  • 0 in reply to HONGYU YAN
    TI__Mastermind 37805 points

    Hi,

    Let me confer internally. I will get back to you tomorrow.

    Best,

    Nate

  • +1
    TI__Intellectual 790 points

    Hello Hongyu.

    You are right. In the fine motion detection mode, the chirps across multiple frames are combined into the radar cube. Hence, this approach results in a non-uniform chirping scheme (since closely spaced chirps within the frames are combined across longer frame durations).

    As a result, as you mentioned, applying the FFT will have artifacts in the generated Doppler spectrum. However, as explained in the tuning guide, this approach is adopted to increase the total chirping window only when detecting fine motions. To achieve this, the Doppler estimation is not directly used in the higher layers (i.e., tracker).

    In normal processing mode (single-frame mode), the Doppler estimations are directly used in the tracker layer. However, when the fine-motion mode runs, the Doppler of the detected points are forced to zero to prevent any confusion in the tracker. In other words, the Doppler estimation is carefully handled in the higher layer blocks. Hence, the artifacts of the non-uniform chirping scheme is minimized in the overall chain.

    Regards.

    Muhammet

  • 0 in reply to Muhammet Yanik
    Prodigy 100 points

    Hi，Muhammet

    Thanks for your reply！I got it！