1 |
|
|
/////////////////////////////////////////////////////////////////////////////// |
2 |
|
|
// BSD 3-Clause License |
3 |
|
|
// |
4 |
|
|
// Copyright (C) 2019-2023, LAAS-CNRS, University of Edinburgh, |
5 |
|
|
// Heriot-Watt University |
6 |
|
|
// Copyright note valid unless otherwise stated in individual files. |
7 |
|
|
// All rights reserved. |
8 |
|
|
/////////////////////////////////////////////////////////////////////////////// |
9 |
|
|
|
10 |
|
|
#include "crocoddyl/core/actions/lqr.hpp" |
11 |
|
|
#include "crocoddyl/core/solvers/fddp.hpp" |
12 |
|
|
#include "crocoddyl/core/states/euclidean.hpp" |
13 |
|
|
#include "crocoddyl/core/utils/callbacks.hpp" |
14 |
|
|
#include "crocoddyl/core/utils/timer.hpp" |
15 |
|
|
|
16 |
|
|
int main(int argc, char* argv[]) { |
17 |
|
|
unsigned int NX = 37; |
18 |
|
|
unsigned int NU = 12; |
19 |
|
|
bool CALLBACKS = false; |
20 |
|
|
unsigned int N = 100; // number of nodes |
21 |
|
|
unsigned int T = 5e3; // number of trials |
22 |
|
|
unsigned int MAXITER = 1; |
23 |
|
|
if (argc > 1) { |
24 |
|
|
T = atoi(argv[1]); |
25 |
|
|
} |
26 |
|
|
|
27 |
|
|
// Creating the action models and warm point for the LQR system |
28 |
|
|
Eigen::VectorXd x0 = Eigen::VectorXd::Zero(NX); |
29 |
|
|
boost::shared_ptr<crocoddyl::ActionModelAbstract> model = |
30 |
|
|
boost::make_shared<crocoddyl::ActionModelLQR>(NX, NU); |
31 |
|
|
std::vector<Eigen::VectorXd> xs(N + 1, x0); |
32 |
|
|
std::vector<Eigen::VectorXd> us(N, Eigen::VectorXd::Zero(NU)); |
33 |
|
|
std::vector<boost::shared_ptr<crocoddyl::ActionModelAbstract> > runningModels( |
34 |
|
|
N, model); |
35 |
|
|
|
36 |
|
|
// Formulating the optimal control problem |
37 |
|
|
boost::shared_ptr<crocoddyl::ShootingProblem> problem = |
38 |
|
|
boost::make_shared<crocoddyl::ShootingProblem>(x0, runningModels, model); |
39 |
|
|
crocoddyl::SolverFDDP solver(problem); |
40 |
|
|
if (CALLBACKS) { |
41 |
|
|
std::vector<boost::shared_ptr<crocoddyl::CallbackAbstract> > cbs; |
42 |
|
|
cbs.push_back(boost::make_shared<crocoddyl::CallbackVerbose>()); |
43 |
|
|
solver.setCallbacks(cbs); |
44 |
|
|
} |
45 |
|
|
|
46 |
|
|
std::cout << "NQ: " |
47 |
|
|
<< solver.get_problem()->get_terminalModel()->get_state()->get_nq() |
48 |
|
|
<< std::endl; |
49 |
|
|
std::cout << "Number of nodes: " << N << std::endl; |
50 |
|
|
|
51 |
|
|
// Solving the optimal control problem |
52 |
|
|
Eigen::ArrayXd duration(T); |
53 |
|
|
for (unsigned int i = 0; i < T; ++i) { |
54 |
|
|
crocoddyl::Timer timer; |
55 |
|
|
solver.solve(xs, us, MAXITER); |
56 |
|
|
duration[i] = timer.get_duration(); |
57 |
|
|
} |
58 |
|
|
|
59 |
|
|
double avrg_duration = duration.sum() / T; |
60 |
|
|
double min_duration = duration.minCoeff(); |
61 |
|
|
double max_duration = duration.maxCoeff(); |
62 |
|
|
std::cout << " FDDP.solve [ms]: " << avrg_duration << " (" << min_duration |
63 |
|
|
<< "-" << max_duration << ")" << std::endl; |
64 |
|
|
|
65 |
|
|
// Running calc |
66 |
|
|
for (unsigned int i = 0; i < T; ++i) { |
67 |
|
|
crocoddyl::Timer timer; |
68 |
|
|
problem->calc(xs, us); |
69 |
|
|
duration[i] = timer.get_duration(); |
70 |
|
|
} |
71 |
|
|
|
72 |
|
|
avrg_duration = duration.sum() / T; |
73 |
|
|
min_duration = duration.minCoeff(); |
74 |
|
|
max_duration = duration.maxCoeff(); |
75 |
|
|
std::cout << " ShootingProblem.calc [ms]: " << avrg_duration << " (" |
76 |
|
|
<< min_duration << "-" << max_duration << ")" << std::endl; |
77 |
|
|
|
78 |
|
|
// Running calcDiff |
79 |
|
|
for (unsigned int i = 0; i < T; ++i) { |
80 |
|
|
crocoddyl::Timer timer; |
81 |
|
|
problem->calcDiff(xs, us); |
82 |
|
|
duration[i] = timer.get_duration(); |
83 |
|
|
} |
84 |
|
|
|
85 |
|
|
avrg_duration = duration.sum() / T; |
86 |
|
|
min_duration = duration.minCoeff(); |
87 |
|
|
max_duration = duration.maxCoeff(); |
88 |
|
|
std::cout << " ShootingProblem.calcDiff [ms]: " << avrg_duration << " (" |
89 |
|
|
<< min_duration << "-" << max_duration << ")" << std::endl; |
90 |
|
|
} |