AWR1843BOOST: Maximum number of point cloud

Hi Kim,

Can you give a breakpoint and check the number of detected points before it stops? Also check if the demo stopped due to increase in number of detected points or any other assert. 

Se Ho Kim said:

What are the limitations of payload size?

It depends on the size of L3 and the UART baudrate to transmit the data over UART.

Se Ho Kim said:

What are the minimum value I can set as thresholds for both range and doppler?

If CFAR threshold decreases, then the number of points and number of false alerts will increase. CFAR threshold is a value from 0-100dB. Based on your configuration, the CFAR threshold should be set.

Regards,

Samhitha

hi, Kim.

I studied the point clould recently, Could you tell you which documents mentioned that the number of points of detected objects is 500,please? thank you.

Se Ho Kim said:

Some documents mentioned the max number of detected points is 50

my detected object is one adult, if I set 1843boost running on 1T4R mode, what's the number of point cloud? then 2T4R mode?

Regards,

Wembanyama

Hi Bobba,

Thank you for a quick response.

Samhitha Bobba said:

Can you give a breakpoint and check the number of detected points before it stops? Also check if the demo stopped due to increase in number of detected points or any other assert.

When I was logging the payload, the number of detObj was less than 100 right before it stopped. Using 921600 Baud rate, as I was the only one in the room and not doing any action in front of radar, I suspect that there are going to be a sudden increase over 100. But I will definitely check if any other assert stopped the radar.

Samhitha Bobba said:

If CFAR threshold decreases, then the number of points and number of false alerts will increase. CFAR threshold is a value from 0-100dB. Based on your configuration, the CFAR threshold should be set.

I see that they can be set from 0 to 100, but 10 and 15 threshold, respectfully, was causing trouble. I am hoping to lower the threshold as much as possible disregarding the false detection. I was curious there are lower bound, because non-zero positive value was causing stoppage on radar. Can I understand this as that 0dB as a minimum, and any value between 0-100 should not have problem?

Regards,

Seho

Hi Wembanyama,

I found the max number of detected points in SDK's document.

ti/mmwave_sdk_03_06_01_00-LTS/packages/ti/demo/xwr18xx/mmw/docs/doxygen/html/index.html

Wembanyama said:

my detected object is one adult, if I set 1843boost running on 1T4R mode, what's the number of point cloud? then 2T4R mode?

I believe that will depend on what your configuration of the radar. You can experiment different configurations and play with variables such as RCS.

https://dev.ti.com/gallery/view/mmwave/mmWave_Demo_Visualizer/ver/3.6.0/

Regards,

Seho

hi Kim,

very appreciate for your help.

Following your prompt, I looked up this information.

Se Ho Kim said:

I found the max number of detected points in SDK's document

My new question is, at excellent resolution(for example 6-8cm resolution), whether the number of targets detected here (Ndmax=500) means the number of point clouds reflected from the target is 500? If I test an adult target with radar 3T4R mode configuration, can the number of dots reach hundreds,not a few points(for example, 5 cloud points,or 8 cloud points)? how to increase the number of cloud points?

and for transmiting the huge cloud point data to PC,  how to transmit these point cloud data (the amount of data seems to be a lot), do you have any ideas or give me some advice,please?

Regards,

Wembanyam

Hi Seho,

Se Ho Kim said:

Can I understand this as that 0dB as a minimum, and any value between 0-100 should not have problem?

Yes, your understanding is correct. This will be valid until the number of detected objects is not greater than the maximum value.

Regards,

Samhitha

Hi Samhitha,

Thank you for the response!

When I measured few more times, it was receiving around 150 points from the radar. On the document, it mentions that the maximum payload is 1 Mbps.

Per TLV packet size would be Frame header(40 bytes), 2 TLV headers for point cloud and side info (8+8), and TLV payload (16 * x, 4 * x). Using 10 Frames per second, around 150 points should not exceed the max payload. On the guimonitor configuration, I am only enabling detected points with side info and using the max Baud rate that it shows on the TI tools. 

If nothing else is enabled but point cloud data and does not exceed UART speed limit, what would be stopping the radar to work with the demo firmware?

I am also curious for the demo firmware whether TI radar stops transmitting if the point cloud exceeds 500 or just cut first 500 points and transmit.

Regards, 

Seho

Hi Wembanyama,

I believe max number of objects, would be maximum number it can get not 1 object as maximum points.

Wembanyama said:

can the number of dots reach hundreds,not a few points(for example, 5 cloud points,or 8 cloud points)? how to increase the number of cloud points?

Number of different configurations could do it. I believe reducing desired RCS, disabling grouping algorithms, or lowering CFAR threshold, would give a lot more number of detected points as I mentioned in original post. However, it would give you false/noise points as well.

Wembanyama said:

and for transmiting the huge cloud point data to PC,  how to transmit these point cloud data (the amount of data seems to be a lot), do you have any ideas or give me some advice,please?

This is also what I am my original post was asking for. When I calculate the payload size, it does not look like I am exceeding the UART limitation, but radar keeps stopping. I am also seeking for help here.

Regards,

Seho

hi Kim,

Se Ho Kim said:

ti/mmwave_sdk_03_06_01_00-LTS/packages/ti/demo/xwr18xx/mmw/docs/doxygen/html/index.html

it mentioned that TLV format data?

I studied today,  maybe the raw ADC data first are processed by MSS and DSS, then there will be an output data "TLV format data",and the board send the TLV data to our PC by UART. if we want 3D point cloud , whether the TLV data means the three-dimensional coordinates (x,y,z), velocity and intensity of the measured point? so what my understand is the size of one TLV data is not large as the raw ADC data, that is about 40 bytes or less? if the size of one TLV data based on processed raw ADC data is 40bytes, then we calculate the size of 500 output data(TLV data) processed by radar MSS and DSS ,transmited by UART, is 20,000 bytes( equals 160,000 bits per sec), it sames close to the boud rate 115200 per sec. right? maybe I'm wrong, give me some details forther, please.

and finally, I  have a question: what demo did you run on your 1843 radar board, and how did you adjust CFAR threshold for as much as point cloud data?

Regards,

Wembanyama

Hi Seho,

Samhitha Bobba said:

Also check if the demo stopped due to increase in number of detected points or any other assert. 

Can you please check when did the demo stop? Is it because of some assert?

Regards,

Samhitha

Hi Samhitha,

How can I check this part? Does it have to be though CCS?

After reaching 200+ detected points, I am not receiving anything from the port checking from visualizer in Python.

I am using the mmw demo from mmwave_sdk_03_06_01_00, I cannot import the project and build from the CCS. I can see there are similar project from radar_toolbox, but not too sure which versions of radar toolbox is compatible with which versions of mmwave sdks.

I am not too sure how can I check this part in CCS.

I would much appreciate if you can provide me more details about ways to check.

Best,

Seho

In case I am supposed to do through CCS, I followed this link for CCS debugging https://dev.ti.com/tirex/explore/node?devtools=AWR1843BOOST&node=A__AKZl2I1OiS5ssBNGpidsCQ__radar_toolbox__1AslXXD__LATEST&placeholder=true

If this is what I am supposed to do so, how can I send the configuration and run the same demo on the debug image? The visualizer cannot be connected, so I cannot run the demo and send the configuration I am using.

Hi Seho,

Today is a holiday in India. I'll get back to you early next week.

Regards,

Samhitha