IWR6843ISK: IWR6843ISK

Hi,

Is it possible that this happens when there is nothing in the field of view so the scaling becomes distorted?

Best,

Nate

Doesn't look like it, there are objects on the scatter plot all the time.

Hi Anna,

Can you replicate the issue reliably? Does it happen under specific conditions? Or randomly in a scene? Can you export the data during that time to see if it looks correct? You could try using the industrial visualizer in the radar toolbox, saving the data (because this visualizer doesn't display range/doppler data) and plot it yourself in 2D to see if the data appears correct.

Best,

Nate

Hi Nate,

The issue is happening persistently, doesn't look like there are any specific conditions are taking place.

The issue is not just in the display of the range-doppler heatmap, because I have already tried recording the data and parsing it later in python and some of the frames were damaged the way I saw them while recording (just the range-doppler map, point cloud of the same frame looks legit). I also tried reading radar's serial directly using python (and not via demo software) and plotting the results myself and still got the same issue... there is only like 1-2% of the frames that are coming corrupted but still would be interesting to know where it's coming from. Can it be the radar itself that outputs the wrong output?

Thank you

Hi Anna, this is possible, but I think it's unlikely that the radar is outputting the wrong data, especially if the point cloud still looks good because the point cloud is based off the range doppler heat map. Can you show a scale on the results so we can be sure that's not the issue? 

You could also look at the detected points in the point cloud and see if they correspond to peaks in what appears to be a corrupted range-doppler heatmap from the same frame. If they do not correspond to peak locations, then I think it's likely this heatmap is either corrupted upon UART transmission or receipt, but if they do correspond to peak locations, then maybe the issue is in the range-doppler heatmap generation. I suspect that the first case is more likely.

Best,

Nate

Right okay, so what I see is the following:

The values in the red box are correspond to the true peaks, they have normal values that usually around 2k-5k, and the the rest of the frame (yellow box) doesn't correspond to anything, it just huge random values between 10k-60k. 

Thank you

Hi Anna,

What code are you running on the device? Is there a reason the range-doppler heatmap might be reading from an incorrect location? Does this happen when you run the out-of-box demo with a configuration provided in the SDK?

Best,

Nate

Hi Nate,

I am using an out-of-the box demo: C:\ti\mmwave_sdk_03_05_00_04\packages\ti\demo\xwr68xx\mmw\xwr68xx_mmw_demo.bin. I tried a different sdk version and unfortunately that didn't help either.

Thank you

Hey, Nate

Haven't heard from you in a while. Any updates?

Thank you

Hi Anna,

What configuration are you using so we can try to replicate?

Best,

Nate

xwr68xx_profile_demo.cfg

Nothing new?

Hi,

Apologies for the delay. I am able to replicate the issue and I will take it to the SDK team to resolve. Please ping this thread in a week if you don't hear back.

Best,

Nate

Thank you for the update.

Any news?

Hi Anna,

Unfortunately I'm still waiting on an answer. Give us another 2 days and I hope to have more information.

Best,

Nate

Hi Nathan,

Thanks for trying. Do you know if this bug will be happening on other mmwave devices as well or is it just IWR6843ISK issue?