Hi 
I’m working on an optimal control problem using the acados_template in Python, and I’ve run into an issue that I’m struggling to resolve. The problem is formulated to control an Markov transition system model with constraints on the control input.
Here are some details about my setup:
- The control input is constrained within a certain range.
- I’m using an
AcadosOcpSolverwith thesqpsolver option. - Even with no constrains the system will get Nan at first step
Unfortunately, when I run my solver, I’m encountering an error with acados returned status 1. I’ve tried to troubleshoot the issue by checking constraints, providing an initial guess, and experimenting with different solver options, but the issue persists.
Here is a snippet of my code where I set up and solve the OCP:
def calculate_motion_mode(temp):
#decompose the rate_matrix
eigenvalues, eigenvectors = np.linalg.eig(get_rate_matrix(temp))
mask = np.abs(eigenvalues) > 1e-5
order = np.argsort(np.abs(eigenvalues[mask]))
return eigenvalues[mask][order], eigenvectors[:, mask][:, order]
def get_rate_matrix(temp):
E = DM([0, 0.4, 1, 0.2])
B = DM([[inf, 1.5, 1.1, inf],
[1.5, inf, 10, 0.01],
[1.1, 10, inf, 1],
[inf, 0.01, 1, inf]])
rate_matrix = exp(-(B-E)/temp).T
# eliminate the inf values
for i in range(4):
for j in range(4):
if B[i, j] == inf:
rate_matrix[i, j] = 0
for j in range(4):
rate_matrix[j, j] = - sum1(rate_matrix[:, j])
return rate_matrix
def get_equilibrium(temp):
rate_matrix = get_rate_matrix(temp)
# rate 2 transition
trans_matrix = rate_matrix/np.max(np.abs(rate_matrix)) + np.eye(4)
# calculate equilibrium
eigenvalues, eigenvectors = np.linalg.eig(trans_matrix)
eigenvectors = np.transpose(eigenvectors)
targetvector = eigenvectors[np.argmax(eigenvalues)]
targetvector = targetvector / np.sum(targetvector)
return targetvector
def export_asm_model() -> AcadosModel:
model_name = 'asm'
# set up states & controls
x1 = SX.sym('x1', 4)
x = vertcat(x1)
T = SX.sym('T')
# xdot
x1_dot = SX.sym('x1_dot', 4)
xdot = vertcat(x1_dot)
# dynamics
f_expl = vertcat(get_rate_matrix(T)@x1)
f_impl = xdot - f_expl
model = AcadosModel()
model.f_impl_expr = f_impl
model.f_expl_expr = f_expl
model.x = x
model.xdot = xdot
model.u = T
model.name = model_name
# set x0 and x_goal
x0 = get_equilibrium(1)
model.x0 = x0
x_goal = get_equilibrium(2)
model.x_goal = x_goal
# motion_mode = calculate_motion_mode(2)
# b = model.x - x_goal
# A = motion_mode[1]
# c = cs.solve(cs.mtimes(A.T, A), cs.mtimes(A.T, b))
# model.con_h_expr_e = vertcat(c[0], c[1])
# model.cost_expr_ext_cost_e = c[2]**2
# store meta information
model.u_labels = ['$T$']
model.t_label = '$t$ [s]'
return model
the solver definition
def main():
Tf = 0.0001
ocp = AcadosOcp()
# set model
model = export_asm_model()
ocp.model = model
nx = model.x.size()[0]
nu = model.u.size()[0]
N = 400
# set number of shooting intervals
ocp.dims.N = N
# set prediction horizon
ocp.solver_options.tf = Tf
# set cost
ocp.constraints.lbu = np.array([1])
ocp.constraints.ubu = np.array([2])
ocp.constraints.idxbu = np.array([0])
ocp.constraints.x0 = model.x0
# set options
ocp.solver_options.qp_solver = 'PARTIAL_CONDENSING_HPIPM' # FULL_CONDENSING_QPOASES
# PARTIAL_CONDENSING_HPIPM, FULL_CONDENSING_QPOASES, FULL_CONDENSING_HPIPM,
# PARTIAL_CONDENSING_QPDUNES, PARTIAL_CONDENSING_OSQP, FULL_CONDENSING_DAQP
ocp.solver_options.hessian_approx = 'GAUSS_NEWTON' # 'GAUSS_NEWTON', 'EXACT'
ocp.solver_options.integrator_type = 'IRK'
# ocp.solver_options.print_level = 1
ocp.solver_options.nlp_solver_type = 'SQP' # SQP_RTI, SQP
ocp.solver_options.globalization = 'MERIT_BACKTRACKING' # turns on globalization
ocp_solver = AcadosOcpSolver(ocp, json_file = 'acados_ocp.json')
simX = np.zeros((N+1, nx))
simU = np.zeros((N, nu))
status = ocp_solver.solve()
error message
iter res_stat res_eq res_ineq res_comp qp_stat qp_iter alpha
0 nan nan 2.000000e+00 0.000000e+00 0 0 0.000000e+00
Traceback (most recent call last):
File "asm_env/asm_ocp.py", line 68, in <module>
main()
File "asm_env/asm_ocp.py", line 54, in main
raise Exception(f'acados returned status {status}.')
Exception: acados returned status 1.
Here are my specific questions:
- What does status code 1 usually indicate in
acados? - Are there any common pitfalls or typical issues that could lead to this error?
- Could someone suggest strategies for further debugging or point out potential issues in the code snippet provided?
I would greatly appreciate any advice or insights from those of you who have experience with acados or similar optimal control problems. If additional information is needed, I am happy to provide it.
Thank you in advance for your time and help!