IWR6843AOPEVM: 3d People Tracking, Increase azimuth range greater than +/- 70 degrees

David Reasoner

Part Number: IWR6843AOPEVM
Other Parts Discussed in Thread: MMWAVE-SDK, IWR6843AOP

I'm using the IWR6843AOPEVM with the MMICWAVEICBOOST with the 3D People Tracking software from the MMWAVE-SDK. Everything is working as expected, but I would like to increase the azimuth range of operation from +/- 70 degrees to +/- 90 degrees.

I made updates in the AOP_9m_default.cfg chirp config file as below for azimuth range of +/- 90 degrees and 1 degree angle search granularity.

fovCfg -1 90.0 24.0

dynamicRangeAngleCfg -1 1.0 0.001 1 0

I see in 3D_people_tracking_detection_layer_tuning_guide.pdf section 3.3.4:

"The current build of the implementation only supports the maximum FOV defined in the table above for the suggested angular search grids. The user needs to adjust the local heap memory to be able to support a wider FOV. If the user configures a wider FOV without changing the mentioned angular search granularities, then it is the user's responsibility to optimize the memory and processing time accordingly."

I believe with my configuration changes shown above that I am ok for memory and processing time since my number of azimuth bins is less than the number of bins using the default values of +/- 70 degrees with 0.75 degree search granularity.

I have gone through the 3D_people_tracking_demo_implementation_guide.pdf document and have an understanding of how the 3D_people_track_6843_dss code is structured. But I cannot locate the code that needs to be changed to enable a larger azimuth field of view.

Please let me know where the azimuth field of view limit is set in the 3D_people_track_6843_dss code.

Thank You!

9 months ago

0 Nathan Block 9 months ago

TI__Mastermind 32400 points

Hi David,

I don't think you'll be able to get +/- 90 degrees of FoV on the IWR6843AOPEVM. The antenna radiation pattern drops down too far in that region.

In fact, the maximum field of view that TI recommends with the IWR6843AOP is +/-60 degrees. It's not a problem that you're using it at +/-70, but I would be quite surprised if you can get any decent performance beyond that.

https://dev.ti.com/tirex/explore/node?a=1AslXXD__1.30.01.03&node=A__AB9Ek2o7Z7KxVGvvfm7HCA__radar_toolbox__1AslXXD__1.30.01.03&r=1AslXXD__1.00.00.26&r=1AslXXD__1.00.01.07

Best,

Nate

0 David Reasoner 9 months ago in reply to Nathan Block

Prodigy 50 points

Nathan, thank you for the reply and the antenna pattern info. I would still like to test out to +/- 90 azimuth. I'm hoping to get at least a few meters range out to +/- 80 degrees. Do you have any information on the code that is limiting the point cloud output TLVs to +/- 70 degrees so that I can modify the code to enable point cloud output to +/- 90 degrees?

0 Nathan Block 9 months ago in reply to David Reasoner

TI__Mastermind 32400 points

Hi,

I'm investigating this now. Give me until EOD monday to get back to you.

In the meantime, can you confirm that this does not detect beyond 70 degrees? Can you turn off clutter removal and use a strong reflector to be sure? Have you done this already?

Best,

Nate

0 David Reasoner 9 months ago in reply to Nathan Block

Prodigy 50 points

Nathan, I've attached a set of point cloud data I collected yesterday. This was taken outdoors in an open field with the EVM mounted on a tripod about 1.5 m high. The target is me walking around at about 2m distance. It appears that the point cloud points are being dropped if they are > 72.8 degress and if they are < -73.3 degrees. Thanks for your continued efforts on this.

dfPointCloud.csv

0 Nathan Block 9 months ago in reply to David Reasoner

TI__Mastermind 32400 points

Thank you David. I'm investigating this now. Give me until EOD monday to get back to you.

Best,

Nate

0 David Reasoner 9 months ago in reply to Nathan Block

Prodigy 50 points

Hi Nathan, any updates on this? Thank You.

0 Nathan Block 9 months ago in reply to David Reasoner

TI__Mastermind 32400 points

Hi David,

Thank you for your patience. Can you put the debugger in the RADARDEMO_aoaEst2DCaponBF_create() function and see if the steering vectors for the range/azimuth heatmap are being generated correctly? The currentAngle variable seems like you should be able to tell if it ranges high enough.

There is also a possibility that this is simply not what the demo is designed to do. You are pushing the field of view beyond the recommended limits, so there is no guarantee that this will fix the underlying issue of no detections.

Best,

Nate

0 David Reasoner 9 months ago in reply to Nathan Block

Prodigy 50 points

Nathan, thanks, I'll test this and get back to you.

0 Nathan Block 9 months ago in reply to David Reasoner

TI__Mastermind 32400 points

Can you also set cfarDiscardLeftAngle and cfarDiscardRightAngle to 0? The default value of 2 will diminish your ability to detect on the edges of the field of view.

Best,

Nate

0 David Reasoner 9 months ago in reply to Nathan Block

Prodigy 50 points

Hi Nathan,

I updated the cfg file to set the discard angles to 0. I also made some other changes to reduce some of the filtering. Here is the dynamicRACfarCfg line in the updated cfg file:

dynamicRACfarCfg -1 4 1 0 0 8 12 2 6 5 3 0.20 1 3

I ran a quick test indoors with the new cfg file, so there is a lot of clutter, as I expected with the changes above. I still see the truncated azimuth angles at +/- 73 degrees. The strong SNR cloud points seen in the plot are from me walking around the sensor at about 2m distance from -90 to +90 degrees azimuth.

I will run in debug mode tomorrow and let you know what I see for the currentAngle variable.

Thanks,

David

3782.dfPointCloud.csv

0 Nathan Block 9 months ago in reply to David Reasoner

TI__Mastermind 32400 points

Thank you David, I look forward to hearing.

Best,

Nate

0 David Reasoner 9 months ago in reply to Nathan Block

Prodigy 50 points

Nathan,

I've been working on monitoring the variables to watch currentAngle and others. I'm successfully running in debug mode in ccs using the procedure in

https://dev.ti.com/tirex/content/radar_toolbox_1_20_00_11/docs/software_guides/using_ccs_debug.html

I also went through mmwave_sdk_user_guide.pdf, specifically section 6.15 on debugging.

I've set breakpoints in the code that are working, and I have the debug window open in CCS that has tabs for "variables", "expressions", "registers", and "breakpoints". However, even when I get to a breakpoint and code execution pauses, the variables window is empty. Can you point me to a document that describes how to monitor the variables?

Thanks,

David

0 Josh Dye 9 months ago in reply to David Reasoner

TI__Genius 16441 points

Hi David,

Nathan is currently out of office. He will get back to you after he returns later this week.

Best Regards,

Josh

0 David Reasoner 9 months ago in reply to Josh Dye

Prodigy 50 points

Ok, thanks.

0 Josh Dye 9 months ago in reply to David Reasoner

TI__Genius 16441 points

Thank you for your patience.

0 Nathan Block 9 months ago in reply to Josh Dye

TI__Mastermind 32400 points

Hi David,

You can look in the expressions window for variables.

Best,

Nate

0 David Reasoner 9 months ago in reply to Nathan Block

Prodigy 50 points

Hi Nathan,

I'm still seeing errors in the expression window - see the snapshot below. I've read multiple e2e forums but have not yet figured out how to see the variables when I hit a breakpoint.

As a work-around, I'm using printf statements to see the value of the variables in the console.

currentAngle ranges from -1.000000 to 1.000003
nuInit is -1.000000
azimSearchLen is 181
fovCfg[0] is 90.000000
nuStep is 0.011111

The azimSearchLen, fovCfg[0] and nuStep values are as I expected given my config file settings for <azimuthFoV> = 90 degrees and <angleSearchStep> = 1 degree.

I'm not sure how to interpret the currentAngle values as they appear to be normalized somewhere else in the code.

Do you have a recommendation for where to look next to identify and update the azimuth angle limitation beyond +/- 73 degrees?

Thanks again for your help on this.

David

0 David Reasoner 9 months ago in reply to David Reasoner

Prodigy 50 points

Nathan,

A quick update - I can now see the variables and expressions in CCS. I had to "Enable Silicon Real-Time Mode". The variable values I'm seeing are consistent with my previous message so my question on the next steps is the same.

Thanks,

David

0 Nathan Block 9 months ago in reply to David Reasoner

TI__Mastermind 32400 points

Hi David,

My next recommendation would be to look into the CFAR detection code. In the RADARDEMO_detectionCFAR_run() you could check to see if detection is possible at the high angle bins. Perhaps they're being skipped in the code? The actual CFAR seems to run in RADARDEMO_detectionCFAR_priv.c

Best,

Nate

0 David Reasoner 8 months ago in reply to Nathan Block

Prodigy 50 points

Hi Nathan,

I collected some additional data outdoors yesterday and observed that at azimuth angles of +70 deg and -70 degrees the radar is not picking up any points even at distances of 5 m and less. I'm seeing very good performance between -50 and +50 degrees. I've concluded that for my application that needs coverage from -90 to +90 degrees I will need to use two radar devices mounted 90 degrees offset, with each covering -45 to +45 degrees.

At this time I'm going to stop working on the effort to expand the software azimuth range of operation.

I have a new question on the distance measurement that I will post in a new thread. Thanks again for your help on this topic.

David

Sensors

Sensors forum

IWR6843AOPEVM: 3d People Tracking, Increase azimuth range greater than +/- 70 degrees