IWR1642: Changes in the detection module for phone detection

Hi Shreekant,

Since you mentioned "high accuracy", I want to clarify the major code implementation of high accuracy lab compared to out-of-box mmw demo. It is based on 1D FFT result. With a detected course peak index as input, zoom-in algorithm generates a fine peak range value. There are no doppler FFT implemented.

Are you planning on making changes on the basis of out-of-box mmw demo or the high accuracy lab? It seems you need to detect and also track the location. The thing is, someone will be holding the phone, right? Is it to track the human or the device only? Why you need high accuracy here?

Thanks and Regards,

Michelle 

Hi,
I need to detect just the phone, and not the human holding the phone.
My experiment is: I’ll put my phone on the table and I need to sense the location of the phone.
I was trying to implement this using the modules like clustering, tracking and range processing inside the “People counting” and “High accuracy” lab experiments.

Hi,

I would suggest to run those labs separately on your experiment setup to see how that works before diving into code update. 

The two labs are quite different, with people counting more complicated. I suggest to start from there if you want to combine. 

Note that anything in the view of the radar sensor reflects radar signal depending on the RCS (radar cross section), for example multi-path can happen with signal reflected by the table surface. 

When you find that the phone detection was OK with your setup, let me know, and we will help to see how to do the code combination as you desired.

Regards,

Michelle 

Hi,

I have already experimented the phone detection with all the Labs provided for IWR1642. But the phone is being considered as a Static clutter. I can’t specifically locate “only” the phone.

I was wondering if I could edit the configuration file, or make some changes to “number of points” inside the clustering module.

That's the result I expected, since the phone is in the middle of all the objects in front of radar EVM. You can hardly pin-point it.
The static clutter will remove all the static points, unless the phone is moving, clutter removal option in the config can't help.

Is the phone moving during test?

Thanks and Regards,
Michelle

Hi,
The experiment is as follows:
There will be a IWR 1642 sensor below the table, and a moving person with a phone in his hand will come towards it, and place the phone on top of the table. I need to find the location once he puts it , i.e. the phone will be static.

Hi,

I'm not sure how the experiment could work. How could the sensor is below the table and then phone is on top of the table? Any objects to be detected needs to be placed un-blocked.

There are two parts, one is moving object -- human with phone and the other is static -- phone placed on the table.

For the moving object detection, people counting can be used -- to detect the human, but I'm not sure if the phone can be detected.
For the static detection, if phone can be placed with flat back facing toward the sensor, reduce the refection of the table by placing the cardboard box on it. Either the oob mmw demo or high accuracy lab can be used. Still setup needs to be carefully placed in order to get a good detection.

Regards,
Michelle

Hi,

Two things:

First:

Even when I perform the people counting lab, the count is not the exact reflection of the actual number. Almost all of the times, it shows more number of people. But in the demonstration videos, it was very accurate.

Second:

I can use the people count lab, for detecting the people coming towards the table. But I’m not so sure how the “High accuracy lab” can be useful for detection of the location of the phone when placed on the table.

Is there some other .bin file in order to configure the sensor? Or can you guys provide me some other code which I can flash?

For your first question, I can re-direct and get expert on people counting to help.

For the 2nd question, the high accuracy lab applies to a stationary object range detection with accuracy range to mm level. It is sensitive to the environment setup. I would recommend using out-of-box mmw demo for a range detection. You can then use visualizer tool to play with the configuration. 

Thanks and Regards,

Michelle

Hi Michelle,

I have performed the lab “high accuracy detection”. I was curious, can we combine the GUI of “people count” and get a similar visualization of the phone, with the help of high accuracy, and somehow find the location with the help of 2D points?