contact-dynamics.cpp
Go to the documentation of this file.
1 //
2 // Copyright (c) 2016-2020 CNRS, INRIA
3 //
4 
5 #include "pinocchio/spatial/se3.hpp"
6 #include "pinocchio/multibody/model.hpp"
7 #include "pinocchio/multibody/data.hpp"
8 #include "pinocchio/algorithm/jacobian.hpp"
9 #include "pinocchio/algorithm/contact-dynamics.hpp"
10 #include "pinocchio/algorithm/joint-configuration.hpp"
11 #include "pinocchio/parsers/sample-models.hpp"
12 #include "pinocchio/utils/timer.hpp"
13 
14 #include <iostream>
15 
16 #include <boost/test/unit_test.hpp>
17 #include <boost/utility/binary.hpp>
18 
19 BOOST_AUTO_TEST_SUITE ( BOOST_TEST_MODULE )
20 
22 {
23  using namespace Eigen;
24  using namespace pinocchio;
25 
28  pinocchio::Data data(model);
29 
30  VectorXd q = VectorXd::Ones(model.nq);
31  q.segment <4> (3).normalize();
32 
33  pinocchio::computeJointJacobians(model, data, q);
34 
35  VectorXd v = VectorXd::Ones(model.nv);
36  VectorXd tau = VectorXd::Zero(model.nv);
37 
38  const std::string RF = "rleg6_joint";
39  const std::string LF = "lleg6_joint";
40 
41  Data::Matrix6x J_RF (6, model.nv);
42  J_RF.setZero();
43  getJointJacobian(model, data, model.getJointId(RF), LOCAL, J_RF);
44  Data::Matrix6x J_LF (6, model.nv);
45  J_LF.setZero();
46  getJointJacobian(model, data, model.getJointId(LF), LOCAL, J_LF);
47 
48  Eigen::MatrixXd J (12, model.nv);
49  J.setZero();
50  J.topRows<6> () = J_RF;
51  J.bottomRows<6> () = J_LF;
52 
53  Eigen::VectorXd gamma (VectorXd::Ones(12));
54 
55  Eigen::MatrixXd H(J.transpose());
56 
57  pinocchio::forwardDynamics(model, data, q, v, tau, J, gamma, 0.);
58  data.M.triangularView<Eigen::StrictlyLower>() = data.M.transpose().triangularView<Eigen::StrictlyLower>();
59 
60  MatrixXd Minv (data.M.inverse());
61  MatrixXd JMinvJt (J * Minv * J.transpose());
62 
63  Eigen::MatrixXd G_ref(J.transpose());
64  cholesky::Uiv(model, data, G_ref);
65  for(int k=0;k<model.nv;++k) G_ref.row(k) /= sqrt(data.D[k]);
66  Eigen::MatrixXd H_ref(G_ref.transpose() * G_ref);
67  BOOST_CHECK(H_ref.isApprox(JMinvJt,1e-12));
68 
69  VectorXd lambda_ref = -JMinvJt.inverse() * (J*Minv*(tau - data.nle) + gamma);
70  BOOST_CHECK(data.lambda_c.isApprox(lambda_ref, 1e-12));
71 
72  VectorXd a_ref = Minv*(tau - data.nle + J.transpose()*lambda_ref);
73 
74  Eigen::VectorXd dynamics_residual_ref (data.M * a_ref + data.nle - tau - J.transpose()*lambda_ref);
75  BOOST_CHECK(dynamics_residual_ref.norm() <= 1e-11); // previously 1e-12, may be due to numerical approximations, i obtain 2.03e-12
76 
77  Eigen::VectorXd constraint_residual (J * data.ddq + gamma);
78  BOOST_CHECK(constraint_residual.norm() <= 1e-12);
79 
80  Eigen::VectorXd dynamics_residual (data.M * data.ddq + data.nle - tau - J.transpose()*data.lambda_c);
81  BOOST_CHECK(dynamics_residual.norm() <= 1e-12);
82 
83 }
84 
85 BOOST_AUTO_TEST_CASE(test_computeKKTMatrix)
86 {
87  using namespace Eigen;
88  using namespace pinocchio;
91  pinocchio::Data data(model), data_ref(model);
92 
93  VectorXd q = VectorXd::Ones(model.nq);
94  q.segment<4>(3).normalize();
95 
96  pinocchio::computeJointJacobians(model, data_ref, q);
97 
98  const std::string RF = "rleg6_joint";
99  const std::string LF = "lleg6_joint";
100 
101  Data::Matrix6x J_RF(6, model.nv);
102  J_RF.setZero();
103  getJointJacobian(model, data_ref, model.getJointId(RF), LOCAL, J_RF);
104  Data::Matrix6x J_LF(6, model.nv);
105  J_LF.setZero();
106  getJointJacobian(model, data_ref, model.getJointId(LF), LOCAL, J_LF);
107 
108  Eigen::MatrixXd J(12, model.nv);
109  J.setZero();
110  J.topRows<6>() = J_RF;
111  J.bottomRows<6>() = J_LF;
112 
113  //Check Forward Dynamics
114  pinocchio::crba(model,data_ref,q);
115  data_ref.M.triangularView<Eigen::StrictlyLower>() = data_ref.M.transpose().triangularView<Eigen::StrictlyLower>();
116 
117  Eigen::MatrixXd MJtJ(model.nv+12, model.nv+12);
118  MJtJ << data_ref.M, J.transpose(),
119  J, Eigen::MatrixXd::Zero(12, 12);
120 
121  Eigen::MatrixXd KKTMatrix_inv(model.nv+12, model.nv+12);
122  computeKKTContactDynamicMatrixInverse(model, data, q, J, KKTMatrix_inv);
123 
124  BOOST_CHECK(KKTMatrix_inv.isApprox(MJtJ.inverse()));
125 }
126 
127 BOOST_AUTO_TEST_CASE(test_getKKTMatrix)
128 {
129  using namespace Eigen;
130  using namespace pinocchio;
133  pinocchio::Data data(model);
134 
135  VectorXd q = VectorXd::Ones(model.nq);
136  q.segment <4> (3).normalize();
137 
138  pinocchio::computeJointJacobians(model, data, q);
139 
140  VectorXd v = VectorXd::Ones(model.nv);
141  VectorXd tau = VectorXd::Zero(model.nv);
142 
143  const std::string RF = "rleg6_joint";
144  const std::string LF = "lleg6_joint";
145 
146  Data::Matrix6x J_RF (6, model.nv);
147  J_RF.setZero();
148  getJointJacobian(model, data, model.getJointId(RF), LOCAL, J_RF);
149  Data::Matrix6x J_LF (6, model.nv);
150  J_LF.setZero();
151  getJointJacobian(model, data, model.getJointId(LF), LOCAL, J_LF);
152 
153  Eigen::MatrixXd J (12, model.nv);
154  J.setZero();
155  J.topRows<6> () = J_RF;
156  J.bottomRows<6> () = J_LF;
157 
158  Eigen::VectorXd gamma (VectorXd::Ones(12));
159 
160  Eigen::MatrixXd H(J.transpose());
161 
162  //Check Forward Dynamics
163  pinocchio::forwardDynamics(model, data, q, v, tau, J, gamma, 0.);
164  data.M.triangularView<Eigen::StrictlyLower>() = data.M.transpose().triangularView<Eigen::StrictlyLower>();
165 
166  Eigen::MatrixXd MJtJ(model.nv+12, model.nv+12);
167  MJtJ << data.M, J.transpose(),
168  J, Eigen::MatrixXd::Zero(12, 12);
169 
170  Eigen::MatrixXd KKTMatrix_inv(model.nv+12, model.nv+12);
171  getKKTContactDynamicMatrixInverse(model, data, J, KKTMatrix_inv);
172 
173  BOOST_CHECK(KKTMatrix_inv.isApprox(MJtJ.inverse()));
174 
175  //Check Impulse Dynamics
176  const double r_coeff = 1.;
177  VectorXd v_before = VectorXd::Ones(model.nv);
178  pinocchio::impulseDynamics(model, data, q, v_before, J, r_coeff, 0.);
179  data.M.triangularView<Eigen::StrictlyLower>() = data.M.transpose().triangularView<Eigen::StrictlyLower>();
180  MJtJ << data.M, J.transpose(),
181  J, Eigen::MatrixXd::Zero(12, 12);
182 
183  getKKTContactDynamicMatrixInverse(model, data, J, KKTMatrix_inv);
184 
185  BOOST_CHECK(KKTMatrix_inv.isApprox(MJtJ.inverse()));
186 }
187 
188 BOOST_AUTO_TEST_CASE ( test_FD_with_damping )
189 {
190  using namespace Eigen;
191  using namespace pinocchio;
192 
195  pinocchio::Data data(model);
196 
197  VectorXd q = VectorXd::Ones(model.nq);
198  q.segment <4> (3).normalize();
199 
200  pinocchio::computeJointJacobians(model, data, q);
201 
202  VectorXd v = VectorXd::Ones(model.nv);
203  VectorXd tau = VectorXd::Zero(model.nv);
204 
205  const std::string RF = "rleg6_joint";
206 
207  Data::Matrix6x J_RF (6, model.nv);
208  J_RF.setZero();
209  getJointJacobian(model, data, model.getJointId(RF), LOCAL, J_RF);
210 
211  Eigen::MatrixXd J (12, model.nv);
212  J.setZero();
213  J.topRows<6> () = J_RF;
214  J.bottomRows<6> () = J_RF;
215 
216  Eigen::VectorXd gamma (VectorXd::Ones(12));
217 
218  // Forward Dynamics with damping
219  pinocchio::forwardDynamics(model, data, q, v, tau, J, gamma, 1e-12);
220 
221  // Matrix Definitions
222  Eigen::MatrixXd H(J.transpose());
223  data.M.triangularView<Eigen::StrictlyLower>() =
224  data.M.transpose().triangularView<Eigen::StrictlyLower>();
225 
226  MatrixXd Minv (data.M.inverse());
227  MatrixXd JMinvJt (J * Minv * J.transpose());
228 
229  // Check that JMinvJt is correctly formed
230  Eigen::MatrixXd G_ref(J.transpose());
231  cholesky::Uiv(model, data, G_ref);
232  for(int k=0;k<model.nv;++k) G_ref.row(k) /= sqrt(data.D[k]);
233  Eigen::MatrixXd H_ref(G_ref.transpose() * G_ref);
234  BOOST_CHECK(H_ref.isApprox(JMinvJt,1e-12));
235 
236  // Actual Residuals
237  Eigen::VectorXd constraint_residual (J * data.ddq + gamma);
238  Eigen::VectorXd dynamics_residual (data.M * data.ddq + data.nle - tau - J.transpose()*data.lambda_c);
239  BOOST_CHECK(constraint_residual.norm() <= 1e-9);
240  BOOST_CHECK(dynamics_residual.norm() <= 1e-12);
241 }
242 
244 {
245  using namespace Eigen;
246  using namespace pinocchio;
247 
250  pinocchio::Data data(model);
251 
252  VectorXd q = VectorXd::Ones(model.nq);
253  q.segment <4> (3).normalize();
254 
255  pinocchio::computeJointJacobians(model, data, q);
256 
257  VectorXd v_before = VectorXd::Ones(model.nv);
258 
259  const std::string RF = "rleg6_joint";
260  const std::string LF = "lleg6_joint";
261 
262  Data::Matrix6x J_RF (6, model.nv);
263  J_RF.setZero();
264  getJointJacobian(model, data, model.getJointId(RF), LOCAL, J_RF);
265  Data::Matrix6x J_LF (6, model.nv);
266  J_LF.setZero();
267  getJointJacobian(model, data, model.getJointId(LF), LOCAL, J_LF);
268 
269  Eigen::MatrixXd J (12, model.nv);
270  J.setZero();
271  J.topRows<6> () = J_RF;
272  J.bottomRows<6> () = J_LF;
273 
274  const double r_coeff = 1.;
275 
276  Eigen::MatrixXd H(J.transpose());
277 
278  pinocchio::impulseDynamics(model, data, q, v_before, J, r_coeff, 0.);
279  data.M.triangularView<Eigen::StrictlyLower>() = data.M.transpose().triangularView<Eigen::StrictlyLower>();
280 
281  MatrixXd Minv (data.M.inverse());
282  MatrixXd JMinvJt (J * Minv * J.transpose());
283 
284  Eigen::MatrixXd G_ref(J.transpose());
285  cholesky::Uiv(model, data, G_ref);
286  for(int k=0;k<model.nv;++k) G_ref.row(k) /= sqrt(data.D[k]);
287  Eigen::MatrixXd H_ref(G_ref.transpose() * G_ref);
288  BOOST_CHECK(H_ref.isApprox(JMinvJt,1e-12));
289 
290  VectorXd lambda_ref = JMinvJt.inverse() * (-r_coeff * J * v_before - J * v_before);
291  BOOST_CHECK(data.impulse_c.isApprox(lambda_ref, 1e-12));
292 
293  VectorXd v_after_ref = Minv*(data.M * v_before + J.transpose()*lambda_ref);
294 
295  Eigen::VectorXd constraint_residual (J * data.dq_after + r_coeff * J * v_before);
296  BOOST_CHECK(constraint_residual.norm() <= 1e-12);
297 
298  Eigen::VectorXd dynamics_residual (data.M * data.dq_after - data.M * v_before - J.transpose()*data.impulse_c);
299  BOOST_CHECK(dynamics_residual.norm() <= 1e-12);
300 
301 }
302 
303 BOOST_AUTO_TEST_CASE (timings_fd_llt)
304 {
305  using namespace Eigen;
306  using namespace pinocchio;
307 
310  pinocchio::Data data(model);
311 
312 #ifdef NDEBUG
313 #ifdef _INTENSE_TESTING_
314  const size_t NBT = 1000*1000;
315 #else
316  const size_t NBT = 100;
317 #endif
318 
319 #else
320  const size_t NBT = 1;
321  std::cout << "(the time score in debug mode is not relevant) " ;
322 #endif // ifndef NDEBUG
323 
324  VectorXd q = VectorXd::Ones(model.nq);
325  q.segment <4> (3).normalize();
326 
327  pinocchio::computeJointJacobians(model, data, q);
328 
329  VectorXd v = VectorXd::Ones(model.nv);
330  VectorXd tau = VectorXd::Zero(model.nv);
331 
332  const std::string RF = "rleg6_joint";
333  const std::string LF = "lleg6_joint";
334 
335  Data::Matrix6x J_RF (6, model.nv);
336  J_RF.setZero();
337  getJointJacobian(model, data, model.getJointId(RF), LOCAL, J_RF);
338  Data::Matrix6x J_LF (6, model.nv);
339  J_LF.setZero();
340  getJointJacobian(model, data, model.getJointId(LF), LOCAL, J_LF);
341 
342  Eigen::MatrixXd J (12, model.nv);
343  J.topRows<6> () = J_RF;
344  J.bottomRows<6> () = J_LF;
345 
346  Eigen::VectorXd gamma (VectorXd::Ones(12));
347 
348  model.lowerPositionLimit.head<7>().fill(-1.);
349  model.upperPositionLimit.head<7>().fill( 1.);
350 
352 
353  PinocchioTicToc timer(PinocchioTicToc::US); timer.tic();
354  SMOOTH(NBT)
355  {
356  pinocchio::forwardDynamics(model, data, q, v, tau, J, gamma, 0.);
357  }
358  timer.toc(std::cout,NBT);
359 
360 }
361 
362 BOOST_AUTO_TEST_SUITE_END ()
double toc()
Definition: timer.hpp:68
void getKKTContactDynamicMatrixInverse(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, const DataTpl< Scalar, Options, JointCollectionTpl > &data, const Eigen::MatrixBase< ConstraintMatrixType > &J, const Eigen::MatrixBase< KKTMatrixType > &KKTMatrix_inv)
Computes the inverse of the KKT matrix for dynamics with contact constraints. It computes the followi...
const DataTpl< Scalar, Options, JointCollectionTpl >::TangentVectorType & impulseDynamics(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, DataTpl< Scalar, Options, JointCollectionTpl > &data, const Eigen::MatrixBase< ConfigVectorType > &q, const Eigen::MatrixBase< TangentVectorType > &v_before, const Eigen::MatrixBase< ConstraintMatrixType > &J, const Scalar r_coeff=0., const Scalar inv_damping=0.)
Compute the impulse dynamics with contact constraints. Internally, pinocchio::crba is called...
const DataTpl< Scalar, Options, JointCollectionTpl >::MatrixXs & crba(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, DataTpl< Scalar, Options, JointCollectionTpl > &data, const Eigen::MatrixBase< ConfigVectorType > &q)
Computes the upper triangular part of the joint space inertia matrix M by using the Composite Rigid B...
q
void computeKKTContactDynamicMatrixInverse(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, DataTpl< Scalar, Options, JointCollectionTpl > &data, const Eigen::MatrixBase< ConfigVectorType > &q, const Eigen::MatrixBase< ConstraintMatrixType > &J, const Eigen::MatrixBase< KKTMatrixType > &KKTMatrix_inv, const Scalar &inv_damping=0.)
Computes the inverse of the KKT matrix for dynamics with contact constraints. It computes the followi...
ConfigVectorType lowerPositionLimit
Lower joint configuration limit.
void randomConfiguration(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, const Eigen::MatrixBase< ConfigVectorIn1 > &lowerLimits, const Eigen::MatrixBase< ConfigVectorIn2 > &upperLimits, const Eigen::MatrixBase< ReturnType > &qout)
Generate a configuration vector uniformly sampled among provided limits.
VectorXs D
Diagonal of the joint space intertia matrix obtained by a Cholesky Decomposition. ...
void getJointJacobian(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, const DataTpl< Scalar, Options, JointCollectionTpl > &data, const typename ModelTpl< Scalar, Options, JointCollectionTpl >::JointIndex jointId, const ReferenceFrame rf, const Eigen::MatrixBase< Matrix6Like > &J)
Computes the Jacobian of a specific joint frame expressed either in the world (rf = WORLD) frame or i...
fill
def H
Definition: dcrba.py:358
void tic()
Definition: timer.hpp:63
VectorXs nle
Vector of Non Linear Effects (dim model.nv). It corresponds to concatenation of the Coriolis...
The LOCAL frame convention corresponds to the frame directly attached to the moving part (Joint...
MatrixXs M
The joint space inertia matrix (a square matrix of dim model.nv).
#define SMOOTH(s)
Definition: timer.hpp:38
TangentVectorType ddq
The joint accelerations computed from ABA.
VectorXs lambda_c
Lagrange Multipliers corresponding to the contact forces in pinocchio::forwardDynamics.
Main pinocchio namespace.
Definition: timings.cpp:28
JointIndex getJointId(const std::string &name) const
Return the index of a joint given by its name.
const DataTpl< Scalar, Options, JointCollectionTpl >::Matrix6x & computeJointJacobians(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, DataTpl< Scalar, Options, JointCollectionTpl > &data, const Eigen::MatrixBase< ConfigVectorType > &q)
Computes the full model Jacobian, i.e. the stack of all motion subspace expressed in the world frame...
TangentVectorType dq_after
Generalized velocity after impact.
int nv
Dimension of the velocity vector space.
Mat & Uiv(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, const DataTpl< Scalar, Options, JointCollectionTpl > &data, const Eigen::MatrixBase< Mat > &v)
Perform the pivot inversion using the Cholesky decomposition stored in data and acting in place...
const DataTpl< Scalar, Options, JointCollectionTpl >::TangentVectorType & forwardDynamics(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, DataTpl< Scalar, Options, JointCollectionTpl > &data, const Eigen::MatrixBase< ConfigVectorType > &q, const Eigen::MatrixBase< TangentVectorType1 > &v, const Eigen::MatrixBase< TangentVectorType2 > &tau, const Eigen::MatrixBase< ConstraintMatrixType > &J, const Eigen::MatrixBase< DriftVectorType > &gamma, const Scalar inv_damping=0.)
Compute the forward dynamics with contact constraints. Internally, pinocchio::computeAllTerms is call...
void normalize(const ModelTpl< Scalar, Options, JointCollectionTpl > &model, const Eigen::MatrixBase< ConfigVectorType > &qout)
Normalize a configuration vector.
ConfigVectorType upperPositionLimit
Upper joint configuration limit.
VectorXs impulse_c
Lagrange Multipliers corresponding to the contact impulses in pinocchio::impulseDynamics.
BOOST_AUTO_TEST_CASE(test_FD)
void humanoidRandom(ModelTpl< Scalar, Options, JointCollectionTpl > &model, bool usingFF=true)
Create a humanoid kinematic tree with 6-DOF limbs and random joint placements.
Eigen::Matrix< Scalar, Eigen::Dynamic, 1 > VectorXd
Definition: conversions.cpp:14
Eigen::Matrix< Scalar, 6, Eigen::Dynamic, Options > Matrix6x
The 6d jacobian type (temporary)
int nq
Dimension of the configuration vector representation.


pinocchio
Author(s):
autogenerated on Fri Jun 23 2023 02:38:29