• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 1075 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.

AWR1642BOOST: AWS1642 radar board raw data capturing

Gelei Deng
Prodigy 100 points
Part Number: AWR1642BOOST

Hi,

I would like to ask about the inbuilt filter within the radar board. Recently I am working on the board and I notice that there is an inbuilt filter in the data processing, which tends to handle the movement of radar.

For instance, I tried to scan the wall with a radar. When I try to turn the radar, the detected objects are rotating but the distance is not changing. This confuses me and gives me much trouble as we tried to integrate the radar on a drone.

So may I know that is there a inbuilt filter within the data capturing? Is it possible for me to turn it off or configure it?

Best regards,

Gelei

  • 0
    Prodigy 100 points

    To explain the problem further, I may want to attach a video demo here.

    We rotate the radar in a constant speed in a square room, and record the data points. As you can see, all the recorded data are in circular. This cannot be the case because at least we expect to detect the square room. I guess that the filter use doppler information to somehow "correct" the moving of radar, and try to show objects on a constant distance. This causes what we see in this video. 

    So could anyone explain on this?

    Best regards,

    Gelei

  • 0 in reply to Gelei Deng
    TI__Mastermind 30405 points
    Hi Gelei,

    There is no such filter in the device or the SDK Demo software (which I assume you are using) which corrects for the movement of Radar to show objects at a constant distance. I would like to get some clarifications regarding your setup as given below.

    1. How is the AWR1642BOOST EVM oriented? My understanding from your point cloud video is that the EVM is rotated 90 degrees i.e. the Azimuth Field of View is being used in the Elevation direction. Is this correct? A picture of the setup which clearly shows the EVM mounting direction would be helpful.

    2. How big is the room and what is the distance of the walls from the mmwave EVM?

    3. I also assume that you have the EVM rotation information being sent to the ROS visualizer which is using it to show and plot the deteced points in the current direction of the EVM (i.e. antenna). Is this correct?

    4. What axis is being used to color code the points on RVIZ display? (i.e. X, Y, Z or intensity?)

    Regards
    -Nitin
  • 0 in reply to Nitin Sakhuja
    Prodigy 100 points

    Dear Nitin,

    Thank you for your replay. For your questions:

    1. Yes you are correct. The radar is rotated 90 degrees. 

    2. The room is quite large, around 10m by 10m. However there are many random objects inside the room. 

    3. Yes sure. Since we do the imaging on linux ROS, we rewrite the axis info and broadcast the detected points in our own node. Visualizer is only used to visualize the generated point cloud map.

    4. It is intensity signal. Actually we don't care much about the intensity value of objects. Currectly we care more about where they are.

    Best regards,

    Gelei

  • 0 in reply to Gelei Deng
    TI__Mastermind 30405 points
    Hi Gelei,

    Thanks for providing the information. This information corroborates our understanding of the visualization shown in the video.

    1. To understand this, you first have to realize that the 1642 EVM does not have any elevation angle resolution. It does have an elevation Field of view of 30 degrees, but all objects anywhere in that elevation FOV will be seen in a 2D plane i.e. a line at 0 degrees.

    2. With the EVM rotated 90 degrees, the elevation FOV becomes Azimuth FOV, so you'll essentially see points in a vertical plane. The points will stay at the same azimuth position even if you rotate the EVM around the the vertical axis as you don't have azimuth angle information in this EVM orientation.

    3. Assuming that the EVM is pointing at 0 degrees in azimuth. In your setup, EVM pose and rotation information is being sent to RVIZ so you'll see points in a line perpendicular to 0 degree azimuth.

    4. Now, when the EVM is rotated by 1 degrees (left or right), RVIZ will plot the vertical line at 1 degree azimuth. It's key to note here that while the visualizer plots a new line at 1 degrees (because of the rotation information sent to it), the points in that line will not change much within the 30 degree FOV, since there's no angle information and all reflections will appear in a straight line.

    5. In effect, points within the EVM azimuth FOV (actually elevation), will appear to be at a constant distance because you're essentially plotting the same/similar points at a different azimuth angle.

    6. You can verify and avoid this effect by plotting the line at larger azimuth steps, say every 15-20 degrees only and you should see the distance of points from center changing.

    Please mark the thread resolved if the above answers your question otherwise let us know if you need more support on this topic.

    Thanks
    -Nitin