Parameteric OCP with path constraint and LS cost

Gaofei · October 6, 2020, 5:51am

@FreyJo @jay101
I also got the same error in matlab interface ,when parameter p needs an update at each stage.

.../casadi/core/sx_function.cpp:147: Code generation is not possible since variables [P_0, P_1] are free.

1.this is an example with nonempty parameter vector:

acados/acados/blob/c63e3ebb437cf1c8fe0b62f37ea22af81f821a5e/examples/acados_python/rsm_example/generate_c_code.py#L119



# set up xdot
psi_d_dot = SX.sym('psi_d_dot')
psi_q_dot = SX.sym('psi_q_dot')
xdot = vertcat(psi_d_dot, psi_q_dot)

# set up parameters
w      = SX.sym('w') # speed
dist_d = SX.sym('dist_d') # d disturbance
dist_q = SX.sym('dist_q') # q disturbance
p      = vertcat(w, dist_d, dist_q)

# build flux expression
Psi = vertcat(psi_d_num(i_d, i_q), psi_q_num(i_d, i_q))

# dynamics
f_impl = vertcat(   psi_d_dot - u_d + Rs*i_d - w*psi_q - dist_d, \
                    psi_q_dot - u_q + Rs*i_q + w*psi_d - dist_q, \
                    psi_d - Psi[0], \
                    psi_q - Psi[1])

2.Here the parameter values are set/updated stagewise:

github.com

acados/acados/blob/c63e3ebb437cf1c8fe0b62f37ea22af81f821a5e/examples/acados_python/rsm_example/generate_c_code.py#L384


    simU[i,j] = u0[j]

field_name = "u"

# update initial condition
x0 = acados_solver.get(1, "x")

if i > Nsim/3 and i < Nsim/2:
    # update params
    for i in range(N):
        acados_solver.set(i, "p", nmp.array([w_val/2.0, 0, 0]))
else:
    # update params
    for i in range(N):
        acados_solver.set(i, "p", nmp.array([w_val, 0, 0]))

acados_solver.set(0, "lbx", x0)
acados_solver.set(0, "ubx", x0)

# update initial condition
x0 = acados_solver.get(1, "x")

I tried the solution you mentioned above . But its not work.

Maybe, its different between Python and Matlab.

So, Are there any examples of MATLAB？

Thank you in advance.

Best，
Gaofei