• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 32 subscribers
  • Views 639 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.

IWR6843AOPEVM: How to extend FOV of Overhead 3D People Counting Lab

rperezti
Genius 12660 points
Part Number: IWR6843AOPEVM
Other Parts Discussed in Thread: IWR6843AOP

Is it possible to extend the FOV of the overhead 3D people counting lab beyond the 4m x 4m square area under the sensor.

We are experimenting with the overhead 3D people counting lab from mmwave_industrial_toolbox_4_4_1 using the IWR6843AOP EVM. By default the demo is configured for an FOV of 4 meters x 4 meters and we would like to extend this to an area of 6 meters x 6 meters. We can see how to modify the visualizer to cover the wider area, but we are trying to understand how to configure the sensor for this. 

We have tried changing the boundary box and static boundary box parameters in the .cfg to the values below but the tracker is only generating outputs within the original 4 x 4 area.

staticBoundaryBox -5 5 -5 5 -0.5 3.25
boundaryBox -6 6 -6 6 -0.5 3.25

Please let us know if it is possible to extend this and how.

Thanks

  • 0
    Intellectual 1000 points

    I remember when running the ISK people counter demo we had to lower the value of profileCfg-->freqSlopeConst to have it see further out.

  • 0 in reply to Mike_d
    TI__Mastermind 28055 points

    Hi Ruben,

    When you are running the demo in the larger area:

    1. What height is the sensor mounted at?
    2. Is the sensor tilted?
    3. Do you see any point cloud output outside of the 4x4 meter area?

    Regards,

    Justin

  • 0 in reply to Justin Curewitz
    Intellectual 1000 points

    Hi Justin,

    The sensor height will be from 2.5 to 3.5 meters

    The sensor may be tilted between 90 and 70 degrees in elevation

    There is no point cloud beyond 3 meters 

    Regards,

    -Mike

  • 0 in reply to Mike_d
    TI__Genius 12660 points

    Justin,

    Just to make sure we're clear on the tilt, when Mike says 90 degrees in elevation the AOP is mounted parallel with the floor. When he says 70 degrees it means the sensor is tilted 20 degrees with respect to the floor in either positive or negative elevation. In all cases, the point cloud is not seen when a person walks beyond 3 meters from the sensor.

    Thanks

  • 0 in reply to rperezti
    TI__Mastermind 30405 points

    Hi Ruben,

    To clarify, the current tracking FoV of the Overhead 3D demo is 3m radial @ 2.8 to 3m mounting height. The inner box is the tracking FoV while the outer boundary box (i.e. static boundary box) is the area where the tracker tries to retain the target for a specified number of frames (instead of de-allocating it immediately), if the target exits the inner boundary box but still remains within the static boundary box.

    It may be not possible to increase the tracking radius of a single sensor to 6m x 6m because of the following constraints

    1. You can do the trigonometric calculation to arrive at the exact FoV and mounting height needed to physically achieve the 6m radius but it will require either a higher mounting height and/or a wider FoV. Also, the search FoV in the demo is limited to 120 degrees because of mirror ghosting and height distortion artifacts seen at the edge of elevation FoV, which is listed in the release notes.

    2. The front-end chirp design in the demo is optimized for 3m mounting height including the transmit power values and will need to be re-designed for higher mounting heights. Without re-designing the complete chirp, you can try to increase the FoV by a small amount by increasing the mounting height (e.g. 3.5m), but you will need to increase the transmit power. The Tx Power Backoff word in the current profileCfg is set to 10dB for each of the 3 TX channels. You will need to make sure that the detection quality is good and produces enough points to meet the required tracker allocation thresholds in the tracker configuration params. 

    Regards

    -Nitin

  • 0 in reply to Nitin Sakhuja
    TI__Genius 12660 points

    Hi Nitin,

    Thank you for the thorough explanation in the last post. I understood this well except for the comment about the inner box and outer boundary box.

    Nitin Sakhuja said:

    To clarify, the current tracking FoV of the Overhead 3D demo is 3m radial @ 2.8 to 3m mounting height. The inner box is the tracking FoV while the outer boundary box (i.e. static boundary box) is the area where the tracker tries to retain the target for a specified number of frames (instead of de-allocating it immediately), if the target exits the inner boundary box but still remains within the static boundary box.

    In the overheard 3D ppl counting lab, the default configuration for boundary boxes are programmed as follows:

    staticBoundaryBox -3 3 -3 3 -0.5 3
    boundaryBox -4 4 -4 4 -0.5 3

    These values suggest to me that the inner box is the static boundary and the outer box is the tracker boundary, but this contradicts your post. Can you please clarify that point further.

    Thanks,

    Ruben

  • 0 in reply to rperezti
    TI__Mastermind 30405 points

    Hi Ruben,

    Yes, you are correct. Sorry for the confusion. The inner box is the static boundary box, while the outer box i.e. boundaryBox defines the overall tracking boundary. The idea is as follows:

    1. boundaryBox: This defines the overall tracking area so only the points which fall within this region are considered by the tracker. Points outside this boundary are discarded.

    2. staticBoundaryBox: This defines the sub-zone of the tracking area, where the tracker tries to maintain tracks for a longer period even if there even if the target does not produce enough detection points to maintain the track. This time period is configured in the stateParam command using the stat2Free argument. This feature is used to maintain tracks for static people say when they are sitting or standing still for a long time without much movement. The targets outside the static boundary box are deleted much quickly once they go out.

    Regards

    -Nitin