Unknown reason for radar detections behind the vehicle

Hi all, I just tried to get radar detections using an automated driving toolbox. I have two questions regarding their usage

The radar setting that I have used for my case is attached in the code below. I have the following questions

1. What are these detections (circled in red) that are behind my vehicle as my radar sensor is only looking forward?
2. I have set the noise to 0. But what should I have to do if I only want to add one type of noise for e.g., Multipath reflections ?
3. Also I would appreciate answer to my previous question https://www.mathworks.com/matlabcentral/answers/1934030-problem-with-the-guardrail-radar-detections-using-automated-driving-toolbox asked about a week before

Could anybody pls answer my questions ?

Thanks in advance for your help

Regards

%generate radar data
radarSensor = drivingRadarDataGenerator( ...
    'SensorIndex', 1, ...%Sensor Identification, ...
    'UpdateRate', 10, ...    
    'MountingLocation', [3.7 0 1.6], ... %Sensor Mounting %20 cm above grd
    'MountingAngles', [0 0 0], ...      % Mounting rotation angles of the radar, in degrees [z_yaw, y_pitch, x_roll]    
    'HasElevation',true, ...%Detection Reporting
    'HasRangeRate',true, ...
    'HasNoise',false,...
    'HasFalseAlarms',0,... %an erroneous radar tgt detn from noise or other interfering signals
    'HasOcclusion', false,... %true results in getting dets only fr radar which has a direct line of sight  
    'MaxNumReportsSource', 'Auto',...%'MaxNumReportsSource', 'Auto', ...%auto - reports all detections or tracks property - only the first N valid dets or tracks, n is defined in 'MaxNumReports' property value
    'TargetReportFormat', 'Detections', ... %tgt reports as unclustered detns, tgt can have multiple detections
    'DetectionCoordinates', 'Sensor spherical', ... 
    'AzimuthResolution', 1, ... %Measurement Resolution  (deg)
    'ElevationResolution', 1, ...    (deg)
    'RangeRateResolution', 0.5, ...(m/s)
    'RangeResolution', 0.25, ...     
    'AzimuthBiasFraction',0.1,... %Measurement Bias     %lower bound on accuracy of azimuth   
    'ElevationBiasFraction', 0.1, ...%
    'RangeBiasFraction', 0.05, ...%
    'RangeRateBiasFraction', 0.05, ...    %
    'DetectionProbability', 0.9, ... %Detection settings      
    'CenterFrequency', 77e9, ... %Hz
    'FalseAlarmRate', [1e-3], ...
    'FieldOfView', [160 160], ...   %deg [azi(0,360] ele (0, 180]
    'RangeLimits', [0 150], ...%m  
    'Profiles', actorProfiles(scenario)); % structure/array that contains the physical and radar characteristics of an actor.

Unrecognized function or variable 'scenario'.

Answer 1

[Serge]

Hi Balaji, my asnwers are as follow:

1. The detections in the red circle should not exist. There is a high chance that the data is not being plotted correctly in this figure. One way to check whether the data you are detecting is within the expected field of view is to evaluate the folowing lines:

% Assign the detections from your radar to a variable called dets. For more info, refer to
% the example in https://www.mathworks.com/help/driving/ug/model-vision-sensor-detections.html

% Estimate the azimuth and elevation angles of your detections
azs = cellfun(@(d)d.Measurement(1),dets);
els = cellfun(@(d)d.Measurement(2),dets);

% Check if all the detections are in your field of view
inFOV = abs(azs)<=rdr.FieldOfView(1)/2 & abs(els)<=rdr.FieldOfView(2)/2
all(inFOV(:))

One caveat here, the estimated azimuth and elevation of the detection can be slightly off due to discretization error/noise.

2. Multipath reflections are not supported in drivingRadarDataGenerator. However, you can use radarDataGenerator to simulate ghost targets as illustrated in https://www.mathworks.com/help/radar/ug/simulating-radar-ghosts-due-to-multipath.html . Additional multipath capabilities should be added for radarDataGenerator in the future releases.

3. If I have some info regarding that thread, I will add it there.