AWR6843AOPEVM: Retrieve UART data from sensor using python code

Update : I partly solved the problem by enabling this section which was commented : (lines 193-195)

Hello Kevin. I didn't know that, thanks for the reply.

Isn't there any TI developped code that could allow me to retrieve the sensor's data ? (preferably python). If my understanding is correct, the mmWave Visualizer is retrieving data from the sensor to output the result, any clues on how that is done ? I'd appreciate to see code snippets if that is possible.

I really need to get the range and velocity to process the rest of my system. 

Thanks and regards,

Lilia

Hello again Kevin,

I have a few questions in the continuity of the radar data recovery unrelated to the code. Please enlighten me.

If I use the detected points TLV (as done in the OOB demo), I get an {x,y,z,velocity} data format : 

1- Is it right to calculate the range of  the detected object with sqrt[x^2+y^2+z^2] ? (since different ways are used in the Range profile TLV)

2- In the mmWave Visualizer version I use, there is always only a 2D scatter plot :

2.1- I'd like to visualize a 3D plot (I read about it in different threads but don't know how to get it). How can I have that ?

2.2- Could you please explain how are the 2D points retrieved ? Is it through a simple projection of the 3D point in the x-y plan ? is this calculation included in the demo code or in the visualizer side ? (This question is justified in 3)

3- I was able to retrieve and decode the sensor's detected points data through another external code : read radar data python sdk3 (I know this is out of TI's support). But there is something I'd like to know about the detected points TLV.

When testing in the mmWave Visualizer, I get for example the (-0.37,1.98) point as seen : 

When I use the code the data is in agreement with the 2D scatter plot X-Y (see related point retreived) :

I also use dynamic objects to check the velocity changes (which guarantees the good recovery of the data) :

The weird thing is that the z parameter is always NULL. As if the radar is sending the detected points TLV with a 2D {x,y} and z=0, instead of a 3D {x,y,z} point, as seen in the 2D scatter plot of the visualizer.

So please, without taking into account the code itself, I want to understand why my radar is sending a 2D data instead of 3D, can this be related to the use of the visualizer ? or is there something to change in the demo code so that the radar also gives the z value.

Hoping to get answers for my request.

Thanks again,

Lilia.

Hi Lilia,

Could you let us know the specific demo and specific visualizer you are using? We have several and I want to make sure we have that info to provide the right support.

Thanks,

Clinton

Hello Clinton,

Thanks for the reply. I am using the following : 

Demo : available in C:\ti\mmwave_sdk_03_06_00_00-LTS\packages\ti\demo\xwr68xx 

mmWave Visualizer : Corresponding to sdk 3.6 version

I thank you for the suggestions regarding the parsing file, however as said before my code is working and getting the same X-Y scatter plot data and right velocity, which mans the data retreiving process is correct.

I hope to know if this data format has anything to do with the radar side (config or code), I will be waiting for any guidance from you.

Thanks a lot !

Lilia.

Hello Kristien,

Excuse my delayed answer. I will split it in 3 parts : 

1) First of all, the .dat file generated is completely incomprehensible (see image bellow).

2) For the time being my problem is still not solved, for example for the obstacle on the image which is at a 10cm distance from the radar, I get the following point on the mmWave Visualizer : (0.3384624 , 0.9246166).

I made sure it is the right point by identifying the detected points before and after adding the object.

If I calculate the range of this object, I get around 0.98m which is absolutely wrong.

I have tested many other objects at different positions and the result is always incoherent. 

3) When using the mentionned code to retreive this data, I get the exact same values as the mmWave Visualizer which means that : 

* The code side can be assumed correct and the troubleshooting is in the radar side.

* The z axis is never taken into account even if the object is not in the X-Y plan.

PS : thank you for your answer in the other thread, effectively it was a matter of endianness and I was able to find the correct structure of the data.

Hoping to get answers.

Thanks,

Lilia.

Hey Lilia,

The .dat file is encoded and cannot be read by a text reader. It would have to be imported back into a JavaScript application using the correct stream decoder to be read properly. For now, I would recommend ignoring my previous recommendation to use the exported data, as it would likely make things more complicated, and it seems that you are retrieving accurate data points. I apologize for making that suggestion. 

As for the incorrect detected point positions, it appears that the antennas are not directed towards the object of interest. For more information on the axes for the antenna, see Section 4.5 xWR6843AOPEVM Antenna of "60GHz mmWave Sensor EVMs" linked below. Figures 4.7 and 4.8 - shown below - should help you orient the antennas properly. 

https://www.ti.com/lit/pdf/swru546 

The lack of z-axis data is likely a result of the configuration used in the demo which may be telling the device to exclude z-axis measurements - though this shouldn't be the case. Could you attach a text file with the console output and save the config you're using and attach that as well? I might be able to determine why the z-axis measurements are missing by parsing through the config file at the very least. 

Regards,

Kristien

Hello Kristien,

First, thank you for keeping up with my request.

I will look at the antennas direction you attached and will be back if I have any questions. Regarding the asked documents, you can find attached the config I am using (I changed the angles to -90/90 for both elevation and azimuth when trying the previous example).

For the output, do you mean the HEX format output or the one where I show my x,y,z,velocity points ? I can give both please confirm only.

Thanks and regards,

Lilia.

 6843AOPconfig10Hz.cfg

Hello Kristien,

I am finally retrieving z data, I'd like to thank you for your assistance.

Your answer resolved the problem.

Thanks and regards,

Lilia