• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 2 answers
  • Subscribers 30 subscribers
  • Views 444 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: Extracting data from simulation

Marvn Arcangel
Genius 12475 points
Part Number: IWR6843ISK
Other Parts Discussed in Thread: IWR6843

Hi Team,

I need help with the use case below.

The IWR6843ISK was used to run the gesture recognition demo successfully. 

A computer simulation model was built through Electromagnetic simulation of the radar scene for the various hand gestures, where you can get the range vs time, velocity vs time, range-doppler profiles etc. 

The question is how can one use this data from the simulations to feed the TI program and extract the features (e.g., Doppler Average, Elevation Weighted Mean, ..., Doppler Azimuth Correlation) mentioned in the link https://dev.ti.com/tirex/explore/node?node=A__ABBjTJ7Y0H4LgRlsbV2S7g__radar_toolbox__1AslXXD__LATEST to finally get the inference for the simulated gestures. How to format the raw simulation data so that TI's program can recognize and able to parse the data just like when it gets the data from the actual radar? Is this even possible?

Thanks!

Regards,

Marvin

  • 0
    TI__Mastermind 20651 points

    Hi Marvin,

    Please help me understand the question a bit more. Currently, all of the processing of the radar data occurs on the IWR6843 device itself. Are you wondering about implementing the same processing and feature extraction on a PC so that the simulation data can be used? Or are you asking if its possible for the device to stream in the data from the simulation and then do its normal processing/feature extraction? 

    Best Regards,

    Josh

  • 0 in reply to Josh Dye
    TI__Genius 12475 points

    Hi Josh,

    Either implementing the same signal processing and feature extraction on a PC (rather than the chip) so that the simulated radar data from the Electromagnetic simulation model (using Ansys HFSS) can be used. Or, if possible for the TI device to stream in the data from the simulation and then do its normal processing/feature extraction, preferably both.

    To start with, the first method would probably be easier as I may just feed a block of data and see the inference is. I still need to figure out how to create steaming data from my simulated results.

    For the machine learning part, it will be good if there's a TI code is available bit I dont think we do. 

    Kindly let me know if either of these is possible, and how to implement.

    Regards,

    Marvin

  • 0 in reply to Marvn Arcangel
    TI__Mastermind 20651 points

    Hi Marvin, 

    I just reached out over email. Please check your inbox. 

    Regards,

    Josh

  • 0 in reply to Josh Dye
    TI__Genius 12475 points

    Hi Josh,

    I sent you an email. have you received it?

  • 0 in reply to Marvn Arcangel
    TI__Mastermind 20651 points

    Hi Marvin, 

    Yes, my apologies. I just replied. 

    Best Regards,

    Josh

  • 0 in reply to Josh Dye
    TI__Genius 12475 points

    Hi Josh,

    No worries, thank you for the update.

    I was wondering if you could point me to a TI documentation that elaborates on the working of the code and especially explanation of the features (may be some mathematical explanation of the features e.g., Doppler Average, Elevation Weighted Mean, Doppler Azimuth Correlation) mentioned in the link dev.ti.com/.../node to extract? 

    Regards,

    Marvin

  • 0 in reply to Marvn Arcangel
    TI__Mastermind 20651 points

    Hi Marvin, 

    We don't currently have any further documentation besides the descriptions in the user guide. You can of course download the Radar Toolbox and view the source code for the demo to gain further understanding. The feature extraction code can be found in gesture.c. 

    Regards,

    Josh