• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 2 answers
  • Subscribers 30 subscribers
  • Views 485 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.

Linux/IWR1443BOOST: Sparse point cloud

Xiaoxuan Lu
Prodigy 130 points
Part Number: IWR1443BOOST
Other Parts Discussed in Thread: IWR1443

Tool/software: Linux

I wonder why the point cloud given by IWR1443 is very sparse on my laptop? I use ROS topic as per the instructions online. 

Also, I want to know to extent this sparsity issue is related to the relatively small number of antennas (3 by 4 array, which is NOT massive MIMO design). Or this is actually limited by the scarce computation resources on board for signal processing? 

Thanks.

  • 0
    TI__Mastermind 33805 points

    Hi,

    Its unclear how you are quantifying the sparsity of the point clouds and what you expect to see.

    is the point cloud similar to that shown in the video linked below?

    https://www.youtube.com/watch?v=0hj9_LKOr3M

    Can you reduce the CFAR and see if that helps?

    Regards,

    Charles O

  • 0 in reply to Charles Oladimeji
    Prodigy 130 points

    Hi Charles,

    Q1: Interesting video but why this board cannot detect and image the wall? Clearly it can detect and image small objects well, but it seems that it needs to be very close to the object (e.g., the glass board on the video) to get dense points for that object. 

    Q2: My point cloud is much sparser, maybe because the object to detect/imagine is wall, rather than small objects. The cfar in my 1843_3d.cfg is:

    cfarCfg -1 0 0 8 4 4 0 768 0
    cfarCfg -1 1 0 4 2 3 1 1280 0

    However, in my experience, if reducing the cfar, the noisy level (SNR) will be worse. I guess that is cost of getting dense point cloud. Could you confirm this?

    Thanks.

  • 0 in reply to Xiaoxuan Lu
    TI__Mastermind 33805 points

    Xiaoxuan Lu said:
    Q1: Interesting video but why this board cannot detect and image the wall? Clearly it can detect and image small objects well, but it seems that it needs to be very close to the object (e.g., the glass board on the video) to get dense points for that object. 

    The walls on the sides are "false"  that the radar can see through, if you watch the video you will notice that towards the end it detected the real red wall at the end,

    Xiaoxuan Lu said:
    My point cloud is much sparser, maybe because the object to detect/imagine is wall, rather than small objects.

    yes there is a trade of between CFAR and whats detected as an object, lower CFAR level picks up more false detection/ noise. You will to fine tune the configurable parameters until you find what works best for your application.

    Regards,

    Charles O 

  • 0 in reply to Charles Oladimeji
    Prodigy 130 points

    Hi Charles,

    Thanks for your reply.

    Regarding the detected wall in the end of the video, it is still not dense enough (compared with its actual width). Is it true?

    For CFAR, now I see why. Because I want filter noises well, so my config does not give me dense point cloud. 

    Best,

    Chris

  • 0 in reply to Xiaoxuan Lu
    TI__Mastermind 33805 points

    Hi,

    The density of the point cloud is not really objectively quantifiable. It really depends on the application.

    If its not dense enough for your application you can adjust the knobs in the configuration file for better performance 

    Regards,

    Charles O

  • 0 in reply to Charles Oladimeji
    Prodigy 130 points

    Having said sparse, I mean compare this mmwave board with Lidar or RGB-D camera.

  • 0 in reply to Xiaoxuan Lu
    TI__Mastermind 33805 points

    You will need a lot more TX, RX antennas for that