• State Resolved
  • Locked Locked
  • Replies 2 replies
  • Subscribers 30 subscribers
  • Views 267 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Original question:

IWR6843ISK: False objects detection on close range OOB demo

IWR6843ISK: Wider-bandwidth chirp impact in false objects detection on close range OOB demo

Rytis Pilipunas
Prodigy 40 points
Part Number: IWR6843ISK

Hi,

I'm performing some test with IWR6843ISK module. I'm using OOB demo from 03_05_00_04 version SDK.

Test scenario:

Radar is stationary. Car driving towards to radar at constant speed, in my case 20 km/h. Visualization of test setup:

Black line is radar FOV and red line is moving car path. Static clutter removal is enabled.

My test results:

While car is in distance greater that ~10m, radar works as expected and detects one object with correct speed. But when car is in closer range radar detects false objects with much lower SNR and wrong speed.

Suggestion from TI was to try wider-bandwidth chirp. So I tried 3 different bandwidths: 0.64 Ghz, 1.6 Ghz and 2.4 Ghz. Originally problem was noticed with 0.64 Ghz bandwidth chirp. I see that wider-bandwidth not help to solve false object detection in close range. 

Added my configuration files, not parsed OOB data, and parsed data in .csv files.

2_20kmh_24ghz_data.csv2_20kmh_16ghz_data.csv2_20kmh_06ghz.cfg2_20kmh_06ghz.dat2_20kmh_24ghz_data.dat2_20kmh_16ghz_data.dat2_20kmh_16ghz_cfg.cfg2_20kmh_24ghz_cfg.cfg2_20kmh_06ghz.csv

Maybe this false detection in close range is just limitation of technology and not much could be done?

Have a nice day,

Rytis

  • +1
    TI__Mastermind 37805 points

    Hi Rytis,

    This data actually looks alright to me when I plot it. It looks like what you're interpreting as noise is multiple reflections coming from different surfaces on the car. This becomes clearer when you plot the data. I looks like you're picking up the side of the car as it gets closer to you, which makes sense because the radar cross section will get larger as you start to see the side of the car too. I would also comment that with the aid of a tracker algorithm, this performance is probably better than it appears.

     

    Best,

    Nate

  • 0 in reply to Nathan Block
    Prodigy 40 points

    Hi Nathan,

    I reviewed my all data several times. In conclusion, I want to agree with Your thoughts, side of the car could be the case in my scenario. Also, I want to mention that CHIRP's with different bandwidth behaves the same.

    Thank You for clear explanation and good customer service.

    Rytis