Struct with cell array to timeseries for Simulink implementation

From the Driving Scenario Designer (Automated Driving Toolbox) I have gathered some radar measurements in the form a struct with the fields Time and ObjectDetections. The ObjectDetections contain a variable number of cells witht the object data. This looks as shown below.

I created this data using Matlab 2021, but for some specific reasons I need to use this data as input in a 2017b Simulink model, but I'm only interested in the measurement data in RadarData_L(:),ObjectDetections(:,1). I tried to convert it to a timeseries object, but then the error 'The first argument must contain the data.' pops up. Should I use timeseries or is there another alternative?

Answer 1

[Rushil]

Hello

I believe that you are using the "drivingScenarioDesigner" function alongside the “drivingRadarDataGenerator” sensor in order to access sensed radar signals for further processing as timeseries data.

The data from RadarData_L can be accessed and converted into timeseries using the following sequence of steps:

• Looping through struct elements for each timestamp
• Keeping track of all timestamps
• Looping through each particular cell array and storing its measurements
• Creating a timeseries object using stored timestamps and measurements

Below is some implementation that may guide you:

len = length(RadarData_L);
times = zeros(len,1);
meas = {}; % cell array to store measurements

for i = 1:len
    times(i) = RadarData_L(i).Time;
    obj_dec = RadarData_L(i).ObjectDetections;
    cur_meas = [];
    
    for j = 1:length(obj_dec)
        det = obj_dec{j}; 
        data = det.Measurement; 
        cur_meas = [cur_meas; data(:)']; % append to prev measurements
    end
    meas{i} = cur_meas;
end

% convert this cell array to a matrix
result = cell2mat(meas);
ans = timeseries(result, times);

Below is the link for “timeseries” that is used above:

https://www.mathworks.com/help/matlab/ref/timeseries.html