from squared signal to sine form

I am working on a project with a robot arm. We made a pathplot with an array in matlab and then integrated this in the simulink model with a "from workspace" block. The problem is that the robot will see squared signals as needing to have a infinit high speed. Therefore, the signal should be rounded off a bit. So it has to get to the value that we gave in in the array but i has to go in a normal speed. Is there a simple way to get this done? Or at least another way than adding a lot of little steps in between?

Thanks!

[Adam Danz]

What are the parameters of the square wave? Are the max and min +/- 1? Is it at a fixed interval? A picture would instantly answer these and other questions.

Answer 1

[Adam Danz]

If the parameters of your square wave are known (amp, phase, frequency, shift), you can probably generate the sine wave directly which would be your best bet.

Another approach is to smooth the sine wave with a Gaussian weighted moving average. That can be done with smoothdata() (r2017a or later).

% Produce & plot an arbitrary square wave
x = -4*pi : .01 : 4*pi; 
sq = ((sin(2*x+4) >=0) - 0.5) * 4; 

plot(x,sq)
ylim([-4,4])

% Smooth it with Gaus moving avg
B = smoothdata(sq,'gaussian'); 

% Show smoothed curve
hold on
plot(x,B,'r-')

[Sabine Havermans]

For some weird reason the smoothdata() function is not working on my data.. The picture attached shows a scope of my array. Is there another option?

[Adam Danz]

When you said you wanted to fit a sin curve to a squared signal, it seems like your signal was a square wave. A sin curve is cyclical so it would be assumed that your square signal is too. But your data are more like a staircase/stairstep. This is why it's always good to share your data or a plot of your data so volunteers here don't spend too much time chasing the wrong interpretation of your problem.

If you attach a mat file with the (x,y) coordinates of your staircase data, I could play around with it. Some ideas off the top of my head are to try a different smoothing algorithm or to fit a sum or sines to the data.

[Sabine Havermans]

I'm sorry for the wrong information, i didn't know the right names for the data. In the attachment there are 4 arrays. Three of these arrays should be smoothened: arrayq1, arrayq2 and arrayR.

Thanks in advance for thinking along with me!

[Adam Danz]

None of your arrays produce the example you shared in scope.JPG.

[Adam Danz]

Based on the data shared in arrays.m, your arrays are [m-by-2] where the first column are the x values and the second column are the y values. You can apply smoothdata() to the y values like this

sd = smoothdata(arrayq1(:,2),'gaussian');

figure
plot(arrayq1(:,1),arrayq1(:,2),'b-','LineWidth',2)
hold on
plot(arrayq1(:,1), sd, 'r-')

[Sabine Havermans]

Unfortunately, this is not an option, because the robot still needs to reach these outliers. I am going to test with the robot if it is working or not. Thank you for helping!