nlmpcmove too many input arguments

Hello, I'm using nlmpc to perform just one prediction step.

I want to use mv = nlmpcmove(nlmpcobj,x,lastmv) from documentation.

I create my nlmpc object like this.

deterministicController = nlmpc(2, 2, 1);
deterministicController.PredictionHorizon = L;

generate a few parameters in a loop and build the prediction models and costs based off of those

deterministicController.Model.StateFcn = @(x,u) epsilons(l) * fNonlinear(x, u, npap) ...
        + (1-epsilons(l)) * fLinear(x, u, npap, xBar);
deterministicController.Model.OutputFcn = @(x,u) x;
deterministicController.Optimization.CustomCostFcn = ...
        @(X, U, data) sum( (...
            [U(1:data.PredictionHorizon,data.MVIndex(1))'; X(2:data.PredictionHorizon+1,1:2)']...
            - zRef) .^2 );

The problem is unconstrained. I then perform just one evaluation with

uStar = nlmpcmove(deterministicController, xm1, um1);

which outputs

Error using nonlinear

Too many input arguments.

Error in znlmpc_getZBounds (line 36)

f1 = hCostFcn(X0, U0, e0, userdata, runtimedata.Parameters{:});

Error in nlmpc/nlmpcmove (line 122)

[zLB, zUB] = znlmpc_getZBounds(coredata, runtimedata, userdata, handles.hCostFcn, z0);

By the way xm1 is 2x1, bm1 is scalar and fLinear, fNonlinear are correct (previously tested, outputting 2x1).

I can provide the rest of the code. Thank you.

Answer 1

[Massimo Pesavento]

I was wrong when setting up the cost function, the form should have been

deterministicController.Optimization.CustomCostFcn = ...
        @(X, U, e, data) sum( (...
            [U(1:data.PredictionHorizon,data.MVIndex(1))'; X(2:data.PredictionHorizon+1,1:2)']...
            - zRef) .^2 );