Hi Champs
I would like to ask if there is a way to correct erroneous detection caused by radar shaking.
The Application is for the rear sensing of the car. and sensing range is btw 5cm ~ 5m
Regards,
Jack
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.
Hi Champs
I would like to ask if there is a way to correct erroneous detection caused by radar shaking.
The Application is for the rear sensing of the car. and sensing range is btw 5cm ~ 5m
Regards,
Jack
Hi,
One more piece of information that may help.
When testing the sensor in mouvement, please make sure that the relative velocity is lower than the Vmax defined by the chirp configuration parameters.
If using the mmWave SDK oob demo, there is an extendedMaxVelocity parameter in the profile configuration file.
If extendedMaxVelocity is enabled it allows to double the Vmax
If the relative velocity is higher than Vmax there will be erroneous detections
thank you
Cesar
Hi Cesar
They use the Traffic Monitoring Lab using Tripod from Outdoor.
When the wind blows outside, a lot of clutter comes up and is detected as a new object, which is mistaken.
In this environment, I would like to ask you if there is a way to correct false detection due to radar shaking instead of mechanical solutions.
Regards,
Jack
Hi Jack,
If the application is for rear ADAS radar systems, what features of the traffic monitoring demo does the customer need that isn't in the automotive demos?
In general, you will want to increase the CFAR thresholds to reduce the number of detections from wind and other large movements. See the SDK user's guide for help with this.
To reduce the number of tracks that appear from wind blowing, you will want to increase the allocationParam thresholds on number or points or SNR for track formation. Please see the following document for info on the allocationParam.
3D_people_counting_tracker_layer_tuning_guide.pdf
Also, if you are only looking up to 5m, make sure the boundaryBox is configured properly to filter out tracks outside that distance.
Regards,
Jackson
Hi Jackson
Thank you for your kind reply.
According to customer contact, they're under develop for traffic monitoring system rather than Car rear sensing project as of now.
Hence they tried to test traffic monitoring lab.
As they tried to apply the above recommendation, In the case of calm winds, there has been some improvement, but there is still a tendency for errors to occur in the case of severe winds, so I would like to ask if there is any further improvement.
1. Is there any additional method other than what you suggested in the thread above? (Very severe wind)
2. Method of improvement if a tree is shaken by the wind and is misdetected (additional questions)
Thanks.
Regards,
Jack
Is the wind shaking the trees and causing the trees to show as detected tracks? Or is the wind blowing and shaking the EVM, causing there to be tracks in random locations? The second case will be difficult to avoid if the sensor is moving.
For other parameters, please see the rest of the document linked above. There are other parameters such as trackingCfg, stateParam, gatingParam that will all affect the tracking.
Regards,
Jackson