IWR6843: People counting and energy reflected based on direction

user5310849

Part Number: IWR6843

Hi TI,

I'm just experimenting with use of people counting demo for detection of fast moving objects such as cyclists. In my configuration, they are passing something like 8m far from radar (that is looking about 30 degrees towards one direction of movement) along the same trajectory. Looking at the pointcloud generated there is huge difference between incoming and receding traffic. As incoming objects can have something like 20+ point, receding ones have just none. Significant moving of CFAR sliders helps for receding objects, but generates too many points for approaching.

Is there any difference of reflected energy based on direction of movement? Or is it some kind of algorithm or configuration issue?

Thank you, Lukas

over 4 years ago

0 Justin Curewitz over 4 years ago

TI__Mastermind 28055 points

Hi Lukas,

Can you give me an estimation of the speed of the target? Can you check the estimated speed against the max velocity of the chirp?

Regards,

Justin

0 user5310849 over 4 years ago in reply to Justin Curewitz

Intellectual 730 points

It could be approx. 10-50pkh. I use default 14m config. In this case it is I guess much more than 23.6kph max stated.

0 Justin Curewitz over 4 years ago in reply to user5310849

TI__Mastermind 28055 points

Hi Lukas,

The Beamforming algorithm we use works better on slow moving objects. Do you see any variation with detection based on speed of the cyclist? Does it perform as expected when a person is walking?

Regards,

Justin

0 user5310849 over 4 years ago in reply to Justin Curewitz

Intellectual 730 points

I works quite good for approaching traffic. Unfortunately, point cloud of receding traffic is much weaker and I'm not sure why.

0 Justin Curewitz over 4 years ago in reply to user5310849

TI__Mastermind 28055 points

Hi Lukas,

Do you always expect that you will be detecting people moving quickly? You may want to modify the chirp configuration to have a higher maximum velocity. If that does not help you may try the long range people detection demo, just modify the chirp slope to have a shorter range and better range resolution.

Regards,

Justin

0 user5310849 over 4 years ago in reply to Justin Curewitz

Intellectual 730 points

Thank you, I will try it. Will the demo (CaponBF, ...) survive even switching from MIMO to non-MIMO (8 to 4 antennas), or is there any additional modification in code or config need?

0 Justin Curewitz over 4 years ago in reply to user5310849

TI__Mastermind 28055 points

Hi Lukas,

The demo is hardcoded to expect 2 Tx 4 Rx Mimo. You will have to modify the device configuration to handle 1 Tx SiMo.

Regards,

Justin

Sensors

Sensors forum

IWR6843: People counting and energy reflected based on direction