AWR1843BOOST: Non-existed objects are detected after changing the start frequency in MRR demo

Hi,

Do you see a difference in behavior if the Start Frequency is 78GHz instead of 78.76Ghz?

Thank you

Cesar

Hi Cesar, 

I use 78.01 instead of 78.76 GHz. Still have this issue. The issue exists for both pruning detected objects and raw detected objects.

Kind regards

Hi,

I would have to talk with the algorithm developer.

This will probably take a few days

thank you

Cesar

Hi,

I have discussed with algorithm developer this issue.

We don't expect to see any difference in behavior if the starting frequency is the only change

Please give us more information about what changes you observe and what is your test environment

thank you

Cesar

Thanks. 

My firmware is base on MRR lab demo. The firmware uses header file "mrr_config_chirp_design_USRR30.h" for USRR configuration.

I only changed below parameter in above header file in both mss and dss's common folder:

#define PROFILE_USRR_START_FREQ_GHZ          (77.01f) 

The comparing parameter setting is:

#define PROFILE_USRR_START_FREQ_GHZ          (78.01f)

I plotted the detected objects of the firmwares with these two different settings as follows. Left is with 77.01f and right with 78.01f. 

I observe that, for the USRR detection, the detected result with setting of 77.01f is very close to reality, while the detected result with setting of 78.01 has big error. 

Hi,

We don't expect to see a difference. What type of environment is this? Is it inside, outside?

Thank you

Cesar

Thanks. I tested inside a warehouse, some obstacles are put in the locations where the detected objects are plotted in the left picture (77.01f). 

Is it available that your team have a test of mrr demo with different starting frequency seetings?

Hi,

We could perform some testing buy only mid-next week.

In the meantime, are you familiar with the mmWave SDK demo?This demo is also based on MIMO and should have similar performance to USRR.

Is it possible to run the SDK 3.5 or 3.6 demo for AWR1843 with the profile

C:\ti\mmwave_sdk_03_05_00_04\packages\ti\demo\xwr18xx\mmw\profiles\profile_2d.cfg

In order to change the starting frequency you can modify the profileCfg

profileCfg 0 77 7 3 39 0 0 100 1 256 7200 0 0 30

profileCfg 0 78 7 3 39 0 0 100 1 256 7200 0 0 30

Thank you

Cesar

Hi Cesar,

Which binary should I use in C:\ti\mmwave_sdk_03_05_00_04\packages\ti\demo\xwr18xx

xwr18xx_mmw_aop_demo.bin

xwr18xx_mmw_demo.bin

Are they the same?

Regards

Hi Cesar,

I tested outbox demo. It doesn't show obvious diffence for using 77 GHz and 78 GHz as the start frequency. 

But in automotive mrr demo. To increase the sensitivity, we disabled the function rangeBasedPruning() in mrr_18xx_dss/dss_data_path.c. 

After this function is disabled, we use the raw detected objects. Then the difference between using 77GHz and 78GHz appears. Ground clutter more frequently appears in the test using 78GHz, but rarely appears in the test using 77GHz.

Regards

Hi,

The MRR demo was optimized for adas applications.The USRR mode was tested with 20-30m ranges and the CFAR thresholds were set experimentally.

If the use case is for range <10m, than the mmWave SDK demo would be better.

If you don't disable rangeBasedPruning(), do you still see a difference between 77GHz and 78Ghz.

I will follow up with the systems team for their input

Thank you

Cesar

Thanks. If I don't disable rangeBasedPruning(), we do not see obvious difference between 77GHz and 78GHz