AWRL6844EVM: cpd

Hi Kevin,

Thank you for reaching out! Please allow us a few days to return to you with some answers.

Best,
Vignesh K.

Of course, thank you.

Hi, Kevin:

Please adjust the zOffsetCfg in the CLI comment for each seat.  

Best,

Zigang

Hello Zigang, 

would you recommend that the seat do not overlay each other? for example the body, footwell and the leg room cube?

Hello Zigang, 

sorry to bombard you with more questions but I am having problems with the sensor properly detecting a car seat facing rearwards. I have adjusted the CFAR threshold K0 and other parameters. I still get no detection of a baby facing rearwards. here is my CFG file:

Hi, Kevin:

Either way is fine (overlap or not overlap).   If you want to skip some region, then you can make them not overlap.  

Best,

Zigang

Hello Zigang,

Wow, you nailed it! most if not all of the problems that I was having are resolved! thank you for your intuition.

on the other hand, there is this odd issue that I get whenever I have an object on the driver's seat. the sensor seems to overpopulate with a bunch of noise only when there is a baby doll on the seat. it's weird because whenever I do not have anything on the seat there is no noise. I would assume that this is because I adjusted the point cloud detection too much.

here is a snippit: 

thank you Zigang.

Hi, Kevin:

Is it possible to let me know which suggestion helps and what is your CFAR setting for now.  Those out of zone point cloud are usually due to the multipath.  For your attached the figure, do you have a baby in the driver seat?  It could be the driver seat causes a lot of multipath that can show in the seat 3.  If you tilting the driver seat down, you may see the detected points in seat 3 reduce dramatically. 

Best,

Zigang

Hey Zigang,

Of course, here is a snippit of the CFG file of my current working settings. The changes that helped in order to distinguish the car seat facing rearwards were creating a bigger box for each seat. I actually changed the X-axis to make it bigger. this seemed to help a lot. I would say that the pre-set zone definition that I originally had wasn't really good.

I see, the baby is sitting in the driver's seat, and I do notice that this causes some noise from what the multipath effect might be. regarding the seat tilt, do you mean tilting the seat back? and would I have to re measure this and implement it to the CFG file?

thank you Zigang

Thank you, Kevin, for the feedback. 

Yes, I mean tilting the seat back. You do not need to modify the configuration file.  Just want you experience how the tilting of the back seat changes the multi-path reflection.   We are still working on improving LPD/CPD.   Hopefully the new algorithm won't be so sensitive to the multi-path reflection.

Best,

Zigang 

By the way, is baby in driver seat or passenger seat a valid test case for this application?  

Best,

Zigang

I would like to say yes. it is something that I have been doing for a while specifically in the footwell of the passenger seat. i'd have to double check.

Hi Zigang,

quick question. when testing the vehicle outside do notice that the wind causes some disturbance? Do you recommend performing test outside?

especially with high winds

here is the euro ncap 2026 regulation

Hi, Kevin:

As I mentioned, we may have some performance gap due to the multipath.   We may have some performance gap with weak baby in strong mind (shaking case).   We are continue to work to improve these cases.

Best,

Zigang