• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 2 answers
  • Subscribers 31 subscribers
  • Views 1307 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.

IWR1642BOOST: mmWave Demo visualizer how to detect a stationary human

Takuma Akimoto
Intellectual 640 points
Part Number: IWR1642BOOST

Hi,

My customer would like to detect a stationary human with mmWave demo visualizer.
I tried that,but it has only detected and plot a moving(walking,running) people at any distance(0~10m).
I use default Best Range Configuration,and "Remove static clutter" was disabled.
Why did this happen with the OOB demo?
Could you tell me the best setting for mmWave Demo visualizer or .cfg file to detect a stationary human?

Best Regards,

Takuma

  • 0
    TI__Mastermind 24955 points
    Takuma,

    I would recommend taking a look at this experiment found on TI's Resource Explorer. The chirp configurations used in this experiment are included: dev.ti.com/.../

    Regards,
    Kyle
  • 0 in reply to Kyle Cousino
    Intellectual 640 points
    Hi, Kyle

    Thank you to your advice. I'm sorry for the late reply.
    I executed OOB demo with the chirp settings you taught andI I was able to confirm that OOB demo can detect human with fine motion.
    My customer try to detect a stationary human at 30m distance.(with using OOB demo).However,the charp configration used in Very fine Detection Demo can only detect 6.4m distance.
    So I tried to change a chirp configuration with using mmWave Sensing Estimator,however I understood it imposible.
    Is it correct in understanding that long-distance sensing can not detect human's fine motions?
    I think People counting demo(with) is more suitable if human sensing is done over long distances, so I might introduce my customer that demo.

    Best Regards,
    Takuma
  • 0 in reply to Takuma Akimoto
    TI__Mastermind 28055 points

    Hi Takuma,

    For the People Counting demo, we have a long range chirp configuration which can be found on this page.  This chirp will have to be modified to detect breathing.  Generally, for detecting small movements, we want to have a very fine velocity resolution, which allows the device to detect the speed of the chest moving as different from the speed of the human.  Increasing velocity resolution will decrease the maximum detectable velocity, and limit the number of frames you can send per second. 

    Can you explain the use case for long range breathing detection? 

    Thanks,

    Justin

  • 0 in reply to Justin Curewitz
    Intellectual 640 points

    Hi Justin,

    When I introduce People counting demo to my customer,  I was planning introduce "Case 5, 50m Outdoor" chirp settings on this page at the same time.

    My customer just began to handle mmWave and they try to test ability of detecting people in long range at mmWave sensor now.

    So they are not very familiar with how people are detected (I am not also too familiar with it…) and they don't have any use case in long range detection. 

    I would like to tell my customer that they need to change the chirp setting and detect minute movements such as heartbeat if they want to detect a stopped person.

    Best Regards,

    Takuma

  • 0
    Intellectual 640 points
    Additionally, I would like to know about person detection in the OOB demo.

    When I created this thread, I asked that OOB demonstration can not detect that human beings are stopped.
    I understand that the OOB demo performs two processes, Range FFT and Doppler FFT, to detect targets.
    If the target speed is zero, we know that it can be detected as Zero Doppler.
    Why OOB demo can not detect stationary people?

    I do not know which document to look at to solve this question.
    I would appreciate it if you could help me.

    Best Regards,

    Takuma
  • 0 in reply to Takuma Akimoto
    TI__Mastermind 28055 points

    Hi Takuma,

    For the People Counting demo, you are correct to have them use a shorter range chirp - I misunderstood the use case. What is the customer trying to accomplish with the mmWave device?   To understand better how People Counting works, please see the TI Design and this paper on tracking and clustering.

    The OOB demo can detect still people, but it has a few settings that can make detection difficult.

    1. Peak Grouping - this feature scans an area of detected points and only shows the strongest. If you are expecting to see many reflections off of a person, ensure this feature is turned off. When using the mmWave Demo Visualizer, this feature can be enabled or disabled in both the range and doppler direction. See the "Group Peaks from Same Object" option.
    2. Static Clutter removal - this removes detection of objects in the 0 doppler bin. It is very useful when looking for movement. However, if you are using a chirp that has relatively poor doppler resolution, small movements or slow movements may not be seen as they will be slow enough to fall into the zero doppler bin. 

    With these features disabled, you will have a richer point cloud. For more insight into these and other configuration parameters, please see the SDK User's Guide, which is available in <SDK_Install_Directory>/docs.  See page 14 for the chirp configuration table.

    You will also need to look into CFAR - I noticed with the default parameters I did not always detect myself.  You will need to lower the CFAR threshold value - this is explained in the User's Guide chirp configuration table as well - see CFARcfg.

    Regards,

    Justin

  • 0 in reply to Justin Curewitz
    Intellectual 640 points
    Hi Justin,

    Thank you for your reply.
    My customers just want to evaluate how much it can do with mmwave and they have not decided on a specific goal.
    So, I will ask customers what they want to achieve ultimately, and then encourage them to make configuration changes based on your advice.

    Best Regards,

    Takuma