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.

IWR1642BOOST: How to use multiple IWR1642 radars in a room for people counting usage

Part Number: IWR1642BOOST
Other Parts Discussed in Thread: IWR1642

Hi Guys,

I want to use multiple IWR1642 radars in a room for people counting. For example, assuming the room is a rectangle, I put radar A at Point (0, 0) facing north, B at (3, 3) facing west, C at (-3, 6) facing southeast. I'll gather the data from every radar in a PC by sockets. My questions are as following:

1. Will the radars interfere with each others, if they are close? If so, how to avoid the interference?

2. Do I need to consider the synchronization between radars?

3. I'll merge the data from radars by coordinate values by the trick of coordinate transform. How to set a universal coordinate for all the radars? I mean, considering the coordinate transform and Azimuth angle together, find a best way to do the coordinate transform simply and not to reduce the accuracy of the radar detection.

Thank you!

Regards,

Michael

  • Hi Michael,

    This ticket has been assigned to the concerned application expert and they will get back to you tomorrow.

    Thanks
    -Nitin
  • Hi Nitin,

    OK, thanks a lot.

    - Michael

  • Hi Michael,

    1. The radars use FMCW design, so they will not cause interference with eachother.
    2. You could synchronize the radar devices so that they are chirping at different times to completely remove the possibility of interference, but this is not necessary for the People Counting use case. If you choose to do this, you will need to calculate the active frame time based on the chirp design, then start the second radar after the first radar has completed its active frame time. For the default People Counting chirp, the active frame time is about 25 ms.
    3. You will want to set 1 radar as the base, which does not coordinate transform. Then you will want to transform the data from the other radar devices. Generally, it is easier to transform in Cartesian space. If you are transforming in polar, you need to know the range and azimuth of the second sensor. Then, you also need to know its rotation with respect to the first sensor.

    Regards,
    Justin
  • Hi Justin,
    Thank you very much. You answer eliminate my many concerns.
    For question 1 and 2, they are clear to me now.
    For question 3, I have the same idea as your your suggestion, but I still doubt what is the best way to config the radar in a corner for coordinate conversion. For example, radar A is at (0, 0) facing north, B, at corner, is at (2, 0) facing northwest. For B, there are two kinds of ways to do the coordinate conversion:
    1. taking azimuth angle as 45 degree, it is just horizontal movement for B's coordinate system respect to A.
    2. taking azimuth angle as 0 degree, it is horizontal movement plus 45 degree rotation for B's coordinate system respect to A.
    Which one is the correct or good (for the accuracy of the result) way?

    (I've tried both, but I can't tell which one is good, as the test environment changes all the time. But the first one is a bit of easy confusing or misunderstanding, considering B may be put at any place with any degree.)

    Regards,
    Michael
  • Hi Michael,

    You should convert the point to cartesian, then do a transformation based on the position of both sensors. Then you can convert back to polar based on the location of the point with respect to the first sensor.

    Regards,
    Justin
  • Hi Justin,
    Yes, I just use the Cartesian coordinate only, never use polar coordinate. My concern is how to consider and handle the radar azimuth angle. As you know, and as it illustrates in the pplcount_user_guide.pdf, when a radar sits in the south and faces northwest,
    1. if the azimuth angle is set to -45 degree, the coordinate of the output track is that X-axis is facing east and Y-axis is facing north.
    2. But, if the azimuth angle is set to 0, the coordinate will be that X-axis is facing northeast and Y-axis is facing northwest.
    So, the setting of azimuth will significantly impact the result and how to transform the result to a base coordinate.
    What do you think about that?

    Another question is, in the above scenario, if set azimuth angle to 0, how to set the SceneryParam? Assuming the room is a 5x5 square, the original point at left-down corner, and radar B at point (4.8, 0.5) facing northwest.

    Regards,
    Michael
  • Hi Justin,
    Yes, I just use the Cartesian coordinate only, never use polar coordinate. My concern is how to consider and handle the radar azimuth angle. As you know, and as it illustrates in the pplcount_user_guide.pdf, when a radar sits in the south and faces northwest,
    1. if the azimuth angle is set to -45 degree, the coordinate of the output track is that X-axis is facing east and Y-axis is facing north.
    2. But, if the azimuth angle is set to 0, the coordinate will be that X-axis is facing northeast and Y-axis is facing northwest.
    So, the setting of azimuth will significantly impact the result and how to transform the result to a base coordinate.
    What do you think about that?

    Another question is, in the above scenario, if set azimuth angle to 0, how to set the SceneryParam? Assuming the room is a 5x5 square, the original point at left-down corner, and radar B at point (4.8, 0.5) facing northwest.

    Regards,
    Michael
  • Hi Michael,

    Since the azimuth angle is made for the visualizer, I'd recommend just setting it to zero in order to make your object data easier to work with an manipulate.

    If you want to discuss your application in more detail then we can possible provide some advice on techniques for manipulating the object data.


    Cheers,
    Akash
  • Hi Akash,
    OK, got it.
    thanks.