Code generation requires variable to be fully assigned variable before subscripting it error in Simulink

4 次查看（过去 30 天）

Gautami Golani 2019-7-8

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/470653-code-generation-requires-variable-to-be-fully-assigned-variable-before-subscripting-it-error-in-simu

评论： Gautami Golani 2019-7-18

Manipulator_program.slx

Dear all,

I am implementing the forward and inverse kinematics of a 6 dof robot arm. I was able to run the code in MATLAB alone and it worked well but when I tried to do the same in SIMULINK, I get an error to define 'theta' and 'dtheta' (the variables in my program) before subscripting it.

Can anyone help me resolve this error?

Appreciate any advice or help

Regards

10 个评论
显示 8更早的评论隐藏 8更早的评论

Walter Roberson 2019-7-8

在 MATLAB Online 中打开

Your current code has

i = 1;
for t=0:tao:T
    stuff
    theta(:,i+1) = theta(:,i) + ((tao*PJ)*(dr -(k*(f_theta - radd)))); 
    theta_dot(:,i)=((tao*PJ)*(dr -(k*(f_theta - radd))));
    i = i + 1;
end

Now 0:tao:T is 1001 entries, so up to theta(:,1002) and theta_dot(:,1001) will be stored into, and i will be left as 1002. Then you have

    f_theta = Trial(theta(:,i));
    
    J = Jacobi(theta(:,i));
    
    PJ = pinv(J);
    
    theta_dot(:,i)=((tao*PJ)*(dr -(k*(f_theta - radd))));

so i there will be 1002 and theta_dot(:,1002) will be stored into, which completes the symmetry of theta(:,1002) already having been stored into.

Then you have

k10 = T/tao+1;

which is k10 = 10/0.01 + 1 so k10 = 1001 .

theta1(1:k10)=theta(1,1:k10);

with theta1 undefined at that point, that would extract the first 1001 columns from theta row 1 and store them into theta1, and likewise for the theta2 to theta6 variables and the dtheta1 to dtheta6 variables. Then

theta = [theta1;theta2;theta3;theta4;theta5;theta6];
dtheta = [dtheta1;dtheta2;dtheta3;dtheta4;dtheta5;dtheta6];

this appears to be reconstructing theta and dtheta, just without the column 1002 that was stored into.

... and then the code ends, without having stored the theta(:,1002) and dtheta(:,1002) into anything. It is not clear as to why you bother to compute dtheta(:,1002) ?

Also, your function appears to be calculating kinematics for time 0 to 10. That would not normally be done in Simulink: normally you would have a current time and a current state and be calculating the next step. The next step could potentially involve forecasting paths, but the output would normally be the single next step, not all of the steps.

Walter Roberson 2019-7-9

You were talking about running the code at the MATLAB level, before starting the Simulink model. You can create the timeseries at the MATLAB level using timeseries()

Gautami Golani 2019-7-18