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: Cloud points of two people walking side by side seen as one cluster rather two

Part Number: IWR1642BOOST
Other Parts Discussed in Thread: IWR6843

Hi there,

I have been evaluating IWR1642Boost for people counting for the application of people counting and tracking inside security entrance control gates used in metro or airport, etc.

I noticed that if two people appear in field of view of mmWave radar side by side with a few centimeters gap (5-15cm gap), radar GUI demo shows one big cluster of cloud points for the two people despite the gap.

I have noticed that the claimed the range resolution is 5cm. I was expecting to see a gap in point cloud due to gap between people. Therefore one track would be made for both of them and this track would be continued all the way through as the get closer to radar which results in wrong people count as one instead of two.

Does this technology has the ability or the potential to separate the people out so that it can give accurate people count? Is there any chip like IWR6843 or AWR series which provide this amount of accuracy?

This is really important for me to know what are the limits and if I can get such accuracy? If this is the case we would be using this kind radar in large scale in our gates and we are multi-national manufacturer for entrace control (

Thanks

Amin

  • Hi Amin,

    Please see the People Counting TI Design for an explanation of how our software works, and the accuracy we have seen with it.

    5cm range resolution refers to the distance at which two different objects can be distinguished by the device.  You can try lowering the range resolution (make the chirp slop smaller), and you will see that many less points are returned per person.  You should also be aware of the angular resolution, which is 15 degrees and cannot be changed.  Two objects need to be 15 degrees apart to be detected by the device.

    The above paragraph is discussing the point cloud, not the tracker output.  The tracker tries to determine where people are by clustering and tracking the radar detection points. After reading the TI Design, please see the customization guide to determine how you may tune the tracker to better separate people who are walking close together.

    Let me know if there is anything I can clarify or dive deeper into.

    Regards,

    Justin