• State Resolved
  • Locked Locked
  • Replies 2 replies
  • Subscribers 30 subscribers
  • Views 3319 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.

IWR1443: Max numbers of objects of the mmWave demo and Q format of xyz

Francois Charlot
Mastermind 28227 points
Part Number: IWR1443

Hi all,

The source code of the mmWave demo defines the maximum number of detected objects to 100 (data_path.h:64 #define MMW_MAX_OBJ_OUT 100).

When a cluttered scene contains more than 100 objects, which logic does the radar sub-system use to select 100 objects? Are they from the first range bins until a number of 100 is reached? Are they the 100 highest peaks as seen by the CFAR function? Other?

A side question is about the Q format of the xyz coordinates of the detected objects. Page 5 of the published data format document (link) tells about a 4-byte descriptor that defines the Q format used for xyz. What is the format of the descriptor itself? E.g., would a description of 0x0008_0a0c correspond to xyz being in format Q8, Q10 and Q12 respectively?

Thank you,
François.

  • 0
    TI__Mastermind 30375 points
    Hello Francois,

    We have contacted the software team to get this information. I will update you when I hear back.

    Regards,

    Adrian
  • 0
    TI__Mastermind 30375 points
    Hello Francois,

    1. CFAR runs along the range bins of the Detection matrix of size RxD where R is number of Range bins and D is number of Doppler bins. Columns 0 to D-1 correspond to Doppler indices 0,1,...,D/2-1,-D/2,-D/2+1,...-1. CFAR scans "active" columns from left to right and stores detected points in the temporary array of size 2K points. Then the peak grouping selects qualified peaks and stores them into the final array of size 100, until it is filled up. In addition AoA can append additional points to the final array, if there is enough space. So there is no algorithm that favors higher peaks, or lower velocity, etc. If there is enough memory, one can modfiy code and increase #define constant MMW_MAX_OBJ_OUT and recompile the code.

    2. The descriptor is a structure containing number of detected points (uint16) and Q format (uint16). There should be a hyper link in the Doxygen document that points to structure definition. In your example the format should be 8, and 80c number of detected objects.

    Regards,

    Adrian