AWR1642BOOST: AWR1642 BOOST : Velocity accuracy of SRR Demo

Hi,

Please give me a few days to get back to you

thank you

Cesar

Hi,

Please respond as soon as possible.

BR

Sreenath

Hi,

Hi,

1) Yes

2) The velocity accuracy is independent of the AoA

Thank you

Cesar

Hi,

In road test, we observed the following :

Predicted velocity of the target vehicle is obtained by relating doppler velocity acquired from mmWave and mmWave mounted vehicle(subject) speed.

It seems that as angle increases the error in velocity of target(vehicle) also increases.

What could be the reason??

BR

Sreenath

Hi,

Please give me a few days to discuss this with our systems team

thank you
Cesar

Hi,

Please revert back as soon as possible.

BR

Sreenath

Hi,

The radar measures the radial velocity. Is the velocity in your table radial velocity?

thank you

Cesar

Hi,

Yes

BR 

Sreenath

Hi,

Radial velocity should not depend on angle information since it is computed only based on doppler fft.

It may be possible in your experiments that the velocity would have a non-radial component.

The non-radial component would be dependent on angle.

This could explain why you observe a dependency on the angle

thank you

Cesar

Hi,

We used the default SRR MATLAB code and stored tracked doppler value( line 600 : detObj.doppler = (detObj.vy.*detObj.y + detObj.vx.*detObj.x)./detObj.range) in csv file

1) Doesn't mmWave provide radial doppler velocity only?

2) What exactly do u mean by non radial component of velocity with respect to mmWave?

3) What could be the reason for mmWave to provide non radial component of doppler velocity?

BR

Sreenath

If you are looking at the tracked speed than this will be angle dependent.

In general, the Doppler velocity directly measured from the radar is the “Radial Velocity”  and hence would  be angle dependent. Note that the radar measurements are in the Spherical coordinates (i.e. Radial Range, Radial Velocity, Azimuth angle).

When the “Actual Target Speed” is measured for their ground truth, it is not in the same co-ordinate system as the radar measurements hence a transformation is required. To measure the velocity in the Cartesian coordinates (x-y plane) you would need to do a transformation based on you specific  geometry ( such as the mounting angle of their sensor and the angle of the vehicle measured from the radar).

Typically, the output of the tracker is in Cartesian co-ordinates (additional correction based on the mount angle of the sensor might still be required for the tracker velocity estimates through).

As I interpret from your table, you are comparing the velocity of the moving vehicle which is not moving in the radial direction. When we “project” the velocity of the vehicle to the Radial direction, we need to incorporate the angle of the vehicle (by a scaling factor of sin(theta)). As the angle estimate can have some small bias/inaccuracies this can potentially introduce some angle dependent errors.  

I think the first to understand is that the velocity v in the information (x,y,z,v)  is the Radial velocity and not the actual velocity of the target and that we need to take into account the mounting angle and angle of the target to convert from Radial velocity to the actual target velocity.

 Thank you

Cesar

Hi, 

Thank you for the detailed explanation.

Does that mean in the above case :

Vr = Vd cos(θ) assuming mounting angle to be zero.

BR

Sreenath

Let me get back to you on Monday

Thank you
Cesar

Hi,

As quoted by TI "the Doppler velocity directly measured from the radar is the “Radial Velocity”  and hence would  be angle dependent". we have observed as angle increases the error in velocity also increases. can you suggest any method to compensate the observed error in velocity?

BR

Sreenath

Hi,

Sorry if this was not clear.

If a vehicle is moving in radial direction, then there is no dependency on the angle.

If the vehicle is moving in non radial direction (which is most of the time the case) then as was mentioned we need to “project” the velocity of the vehicle to the Radial direction and we need to incorporate the angle of the vehicle

Thank you

Cesar

Hi,

Even after we project the velocity of vehicle to radial direction ,  the error still persists.

what could be the solution?

BR

Sreenath