AWR1642BOOST: Questions about moving human detection

Edward Lee3

Part Number: AWR1642BOOST
Other Parts Discussed in Thread: AWR1642

Hi,

we tried to use the AWR1642 EVM and mmWave demo visualizer to detect a human moving towards and away from the radar along the
DOA = 0 degree direction.

We got multiple detection points which may be considered as two clusters:

The two clusters are distributed around the left side and the right side of the moving person;
it seems that the cluster distribution are symmetric about the person.

The farther the distance between the radar and the moving person is, the larger the separation between the two clusters is.

The separation between the two clusters is larger than 3 meters when the distance between the radar and the person is larger
than 7.5 meter according the "X-Y Scatter Plot" shown in mmWave demo visualizer.

Our questions:
1. Why does the radar return such detection result?
2. Are there any parameters we could adjust to get correct detection point?

Thanks for your help.

Edward Lee

our radar settings:

detection result:

over 4 years ago

0 Cesar over 4 years ago

TI__Guru*** 147894 points

Hi,

This could be multi-path reflections. If you repeat the experiment in an open environment do you still see the 2 clusters?

Thank you

Cesar

0 Edward Lee3 over 4 years ago in reply to Cesar

Intellectual 260 points

Hi,

actually we conducted this experiment in an open environment as you could see in the image below.
Could this be multi-path reflections?

Thanks for your help.

Edward Lee

0 Cesar over 4 years ago in reply to Edward Lee3

TI__Guru*** 147894 points

No,

This does not seem to be multi-path reflections.

We would need more information to debug.

Are there any tests where you don't get this behavior?

Was this behavior always the same or did it start at some point? Maybe it is an EVM issue.

Thank you

Cesar

0 Edward Lee3 over 4 years ago in reply to Cesar

Intellectual 260 points

Hi,

we conducted other experiments:

x-direction movement: A person moved from the left to the right, then from the right to the left

The detection points are much closer than the case we mentioned in the above.

Please check the video: x_moving.avi

radial movement: A person moved towards and away from the radar

We observed the similar results that the separation between the clusters is large.

Please check the videos:
DOA_01.avi: positive DOA
DOA_02.avi: negative DOA

move-stop-move: A person moved towards and away from the radar along the DOA = 0 degree direction; the person stopped few times during this experiment

The person stopped at t = 4-5 sec., 9-11 sec., 15-16 sec., 21 sec.
We observed that the detection points are more convergent when the person stopped moving;
when the person started moving and we got divergent clusters again.

Please check the video: move_stop_move.avi

Thanks for your help.

Edward Lee