| Line |
Branch |
Exec |
Source |
| 1 |
|
|
// |
| 2 |
|
|
// Copyright (c) 2019 INRIA |
| 3 |
|
|
// |
| 4 |
|
|
|
| 5 |
|
|
#include "pinocchio/algorithm/jacobian.hpp" |
| 6 |
|
|
#include "pinocchio/algorithm/rnea.hpp" |
| 7 |
|
|
#include "pinocchio/algorithm/rnea-derivatives.hpp" |
| 8 |
|
|
#include "pinocchio/algorithm/kinematics-derivatives.hpp" |
| 9 |
|
|
#include "pinocchio/algorithm/contact-dynamics.hpp" |
| 10 |
|
|
#include "pinocchio/algorithm/joint-configuration.hpp" |
| 11 |
|
|
#include "pinocchio/multibody/sample-models.hpp" |
| 12 |
|
|
|
| 13 |
|
|
#include <iostream> |
| 14 |
|
|
|
| 15 |
|
|
#include <boost/test/unit_test.hpp> |
| 16 |
|
|
#include <boost/utility/binary.hpp> |
| 17 |
|
|
|
| 18 |
|
|
BOOST_AUTO_TEST_SUITE(BOOST_TEST_MODULE) |
| 19 |
|
|
|
| 20 |
|
|
using namespace Eigen; |
| 21 |
|
|
using namespace pinocchio; |
| 22 |
|
|
|
| 23 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_FD_with_contact_cst_gamma) |
| 24 |
|
|
{ |
| 25 |
|
✗ |
pinocchio::Model model; |
| 26 |
|
✗ |
pinocchio::buildModels::humanoidRandom(model, true); |
| 27 |
|
✗ |
pinocchio::Data data(model), data_check(model); |
| 28 |
|
|
|
| 29 |
|
✗ |
VectorXd q = VectorXd::Ones(model.nq); |
| 30 |
|
✗ |
normalize(model, q); |
| 31 |
|
|
|
| 32 |
|
✗ |
computeJointJacobians(model, data, q); |
| 33 |
|
|
|
| 34 |
|
✗ |
VectorXd v = VectorXd::Ones(model.nv); |
| 35 |
|
✗ |
VectorXd tau = VectorXd::Random(model.nv); |
| 36 |
|
|
|
| 37 |
|
✗ |
const std::string RF = "rleg6_joint"; |
| 38 |
|
✗ |
const Model::JointIndex RF_id = model.getJointId(RF); |
| 39 |
|
|
// const std::string LF = "lleg6_joint"; |
| 40 |
|
|
// const Model::JointIndex LF_id = model.getJointId(LF); |
| 41 |
|
|
|
| 42 |
|
✗ |
Data::Matrix6x J_RF(6, model.nv); |
| 43 |
|
✗ |
J_RF.setZero(); |
| 44 |
|
✗ |
getJointJacobian(model, data, RF_id, LOCAL, J_RF); |
| 45 |
|
✗ |
Motion::Vector6 gamma_RF; |
| 46 |
|
✗ |
gamma_RF.setZero(); |
| 47 |
|
✗ |
forwardKinematics(model, data, q, v, VectorXd::Zero(model.nv)); |
| 48 |
|
✗ |
gamma_RF += data.a[RF_id].toVector(); // Jdot * qdot |
| 49 |
|
|
|
| 50 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_PUSH |
| 51 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_IGNORED_DEPRECECATED_DECLARATIONS |
| 52 |
|
✗ |
forwardDynamics(model, data, q, v, tau, J_RF, gamma_RF); |
| 53 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_POP |
| 54 |
|
|
|
| 55 |
|
✗ |
VectorXd ddq_ref = data.ddq; |
| 56 |
|
✗ |
Force::Vector6 contact_force_ref = data.lambda_c; |
| 57 |
|
|
|
| 58 |
|
✗ |
container::aligned_vector<Force> fext((size_t)model.njoints, Force::Zero()); |
| 59 |
|
✗ |
fext[RF_id] = ForceRef<Force::Vector6>(contact_force_ref); |
| 60 |
|
|
|
| 61 |
|
|
// check call to RNEA |
| 62 |
|
✗ |
rnea(model, data_check, q, v, ddq_ref, fext); |
| 63 |
|
|
|
| 64 |
|
✗ |
BOOST_CHECK(data_check.tau.isApprox(tau)); |
| 65 |
|
✗ |
forwardKinematics(model, data_check, q, VectorXd::Zero(model.nv), ddq_ref); |
| 66 |
|
✗ |
BOOST_CHECK(data_check.a[RF_id].toVector().isApprox(-gamma_RF)); |
| 67 |
|
|
|
| 68 |
|
✗ |
Data data_fd(model); |
| 69 |
|
✗ |
VectorXd q_plus(model.nq); |
| 70 |
|
✗ |
VectorXd v_eps(model.nv); |
| 71 |
|
✗ |
v_eps.setZero(); |
| 72 |
|
✗ |
VectorXd v_plus(v); |
| 73 |
|
✗ |
VectorXd tau_plus(tau); |
| 74 |
|
✗ |
const double eps = 1e-8; |
| 75 |
|
|
|
| 76 |
|
|
// check: dddq_dtau and dlambda_dtau |
| 77 |
|
✗ |
MatrixXd dddq_dtau(model.nv, model.nv); |
| 78 |
|
✗ |
Data::Matrix6x dlambda_dtau(6, model.nv); |
| 79 |
|
|
|
| 80 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 81 |
|
|
{ |
| 82 |
|
✗ |
tau_plus[k] += eps; |
| 83 |
|
|
|
| 84 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_PUSH |
| 85 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_IGNORED_DEPRECECATED_DECLARATIONS |
| 86 |
|
✗ |
forwardDynamics(model, data_fd, q, v, tau_plus, J_RF, gamma_RF); |
| 87 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_POP |
| 88 |
|
|
|
| 89 |
|
✗ |
const Data::TangentVectorType & ddq_plus = data_fd.ddq; |
| 90 |
|
✗ |
Force::Vector6 contact_force_plus = data_fd.lambda_c; |
| 91 |
|
|
|
| 92 |
|
✗ |
dddq_dtau.col(k) = (ddq_plus - ddq_ref) / eps; |
| 93 |
|
✗ |
dlambda_dtau.col(k) = (contact_force_plus - contact_force_ref) / eps; |
| 94 |
|
|
|
| 95 |
|
✗ |
tau_plus[k] -= eps; |
| 96 |
|
|
} |
| 97 |
|
|
|
| 98 |
|
✗ |
MatrixXd A(model.nv + 6, model.nv + 6); |
| 99 |
|
✗ |
data.M.transpose().triangularView<Eigen::Upper>() = data.M.triangularView<Eigen::Upper>(); |
| 100 |
|
✗ |
A.topLeftCorner(model.nv, model.nv) = data.M; |
| 101 |
|
✗ |
A.bottomLeftCorner(6, model.nv) = J_RF; |
| 102 |
|
✗ |
A.topRightCorner(model.nv, 6) = J_RF.transpose(); |
| 103 |
|
✗ |
A.bottomRightCorner(6, 6).setZero(); |
| 104 |
|
|
|
| 105 |
|
✗ |
MatrixXd Ainv = A.inverse(); |
| 106 |
|
✗ |
BOOST_CHECK(Ainv.topRows(model.nv).leftCols(model.nv).isApprox(dddq_dtau, std::sqrt(eps))); |
| 107 |
|
✗ |
BOOST_CHECK(Ainv.bottomRows(6).leftCols(model.nv).isApprox(-dlambda_dtau, std::sqrt(eps))); |
| 108 |
|
|
|
| 109 |
|
|
// check: dddq_dv and dlambda_dv |
| 110 |
|
✗ |
MatrixXd dddq_dv(model.nv, model.nv); |
| 111 |
|
✗ |
Data::Matrix6x dlambda_dv(6, model.nv); |
| 112 |
|
|
|
| 113 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 114 |
|
|
{ |
| 115 |
|
✗ |
v_plus[k] += eps; |
| 116 |
|
|
|
| 117 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_PUSH |
| 118 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_IGNORED_DEPRECECATED_DECLARATIONS |
| 119 |
|
✗ |
forwardDynamics(model, data_fd, q, v_plus, tau, J_RF, gamma_RF); |
| 120 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_POP |
| 121 |
|
|
|
| 122 |
|
✗ |
const Data::TangentVectorType & ddq_plus = data_fd.ddq; |
| 123 |
|
✗ |
Force::Vector6 contact_force_plus = data_fd.lambda_c; |
| 124 |
|
|
|
| 125 |
|
✗ |
dddq_dv.col(k) = (ddq_plus - ddq_ref) / eps; |
| 126 |
|
✗ |
dlambda_dv.col(k) = (contact_force_plus - contact_force_ref) / eps; |
| 127 |
|
|
|
| 128 |
|
✗ |
v_plus[k] -= eps; |
| 129 |
|
|
} |
| 130 |
|
|
|
| 131 |
|
✗ |
computeRNEADerivatives(model, data_check, q, v, VectorXd::Zero(model.nv)); |
| 132 |
|
✗ |
MatrixXd dddq_dv_anal = -Ainv.topRows(model.nv).leftCols(model.nv) * data_check.dtau_dv; |
| 133 |
|
✗ |
MatrixXd dlambda_dv_anal = -Ainv.bottomRows(6).leftCols(model.nv) * data_check.dtau_dv; |
| 134 |
|
|
|
| 135 |
|
✗ |
BOOST_CHECK(dddq_dv_anal.isApprox(dddq_dv, std::sqrt(eps))); |
| 136 |
|
✗ |
BOOST_CHECK(dlambda_dv_anal.isApprox(-dlambda_dv, std::sqrt(eps))); |
| 137 |
|
|
|
| 138 |
|
✗ |
MatrixXd dddq_dq(model.nv, model.nv); |
| 139 |
|
✗ |
Data::Matrix6x dlambda_dq(6, model.nv); |
| 140 |
|
|
|
| 141 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 142 |
|
|
{ |
| 143 |
|
✗ |
v_eps[k] = eps; |
| 144 |
|
✗ |
q_plus = integrate(model, q, v_eps); |
| 145 |
|
✗ |
computeJointJacobians(model, data_fd, q_plus); |
| 146 |
|
✗ |
getJointJacobian(model, data_fd, RF_id, LOCAL, J_RF); |
| 147 |
|
|
|
| 148 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_PUSH |
| 149 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_IGNORED_DEPRECECATED_DECLARATIONS |
| 150 |
|
✗ |
forwardDynamics(model, data_fd, q_plus, v, tau, J_RF, gamma_RF); |
| 151 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_POP |
| 152 |
|
|
|
| 153 |
|
✗ |
const Data::TangentVectorType & ddq_plus = data_fd.ddq; |
| 154 |
|
✗ |
Force::Vector6 contact_force_plus = data_fd.lambda_c; |
| 155 |
|
|
|
| 156 |
|
✗ |
dddq_dq.col(k) = (ddq_plus - ddq_ref) / eps; |
| 157 |
|
✗ |
dlambda_dq.col(k) = (contact_force_plus - contact_force_ref) / eps; |
| 158 |
|
|
|
| 159 |
|
✗ |
v_eps[k] = 0.; |
| 160 |
|
|
} |
| 161 |
|
|
|
| 162 |
|
✗ |
computeRNEADerivatives(model, data_check, q, v, ddq_ref, fext); |
| 163 |
|
✗ |
Data::Matrix6x v_partial_dq(6, model.nv), a_partial_dq(6, model.nv), a_partial_dv(6, model.nv), |
| 164 |
|
✗ |
a_partial_da(6, model.nv); |
| 165 |
|
✗ |
v_partial_dq.setZero(); |
| 166 |
|
✗ |
a_partial_dq.setZero(); |
| 167 |
|
✗ |
a_partial_dv.setZero(); |
| 168 |
|
✗ |
a_partial_da.setZero(); |
| 169 |
|
✗ |
Data data_kin(model); |
| 170 |
|
✗ |
computeForwardKinematicsDerivatives(model, data_kin, q, VectorXd::Zero(model.nv), ddq_ref); |
| 171 |
|
✗ |
getJointAccelerationDerivatives( |
| 172 |
|
|
model, data_kin, RF_id, LOCAL, v_partial_dq, a_partial_dq, a_partial_dv, a_partial_da); |
| 173 |
|
|
|
| 174 |
|
✗ |
MatrixXd dddq_dq_anal = -Ainv.topRows(model.nv).leftCols(model.nv) * data_check.dtau_dq; |
| 175 |
|
✗ |
dddq_dq_anal -= Ainv.topRows(model.nv).rightCols(6) * a_partial_dq; |
| 176 |
|
|
|
| 177 |
|
✗ |
MatrixXd dlambda_dq_anal = Ainv.bottomRows(6).leftCols(model.nv) * data_check.dtau_dq; |
| 178 |
|
✗ |
dlambda_dq_anal += Ainv.bottomRows(6).rightCols(6) * a_partial_dq; |
| 179 |
|
|
|
| 180 |
|
✗ |
BOOST_CHECK(dddq_dq_anal.isApprox(dddq_dq, std::sqrt(eps))); |
| 181 |
|
✗ |
BOOST_CHECK(dlambda_dq_anal.isApprox(dlambda_dq, std::sqrt(eps))); |
| 182 |
|
|
} |
| 183 |
|
|
|
| 184 |
|
|
template<typename ConfigVectorType, typename TangentVectorType1, typename TangentVectorType2> |
| 185 |
|
✗ |
VectorXd constraintDynamics( |
| 186 |
|
|
const Model & model, |
| 187 |
|
|
Data & data, |
| 188 |
|
|
const Eigen::MatrixBase<ConfigVectorType> & q, |
| 189 |
|
|
const Eigen::MatrixBase<TangentVectorType1> & v, |
| 190 |
|
|
const Eigen::MatrixBase<TangentVectorType2> & tau, |
| 191 |
|
|
const Model::JointIndex id) |
| 192 |
|
|
{ |
| 193 |
|
✗ |
computeJointJacobians(model, data, q); |
| 194 |
|
✗ |
Data::Matrix6x J(6, model.nv); |
| 195 |
|
✗ |
J.setZero(); |
| 196 |
|
|
|
| 197 |
|
✗ |
getJointJacobian(model, data, id, LOCAL, J); |
| 198 |
|
✗ |
Motion::Vector6 gamma; |
| 199 |
|
✗ |
forwardKinematics(model, data, q, v, VectorXd::Zero(model.nv)); |
| 200 |
|
✗ |
gamma = data.a[id].toVector(); |
| 201 |
|
|
|
| 202 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_PUSH |
| 203 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_IGNORED_DEPRECECATED_DECLARATIONS |
| 204 |
|
✗ |
forwardDynamics(model, data, q, v, tau, J, gamma); |
| 205 |
|
|
PINOCCHIO_COMPILER_DIAGNOSTIC_POP |
| 206 |
|
|
|
| 207 |
|
✗ |
VectorXd res(VectorXd::Zero(model.nv + 6)); |
| 208 |
|
|
|
| 209 |
|
✗ |
res.head(model.nv) = data.ddq; |
| 210 |
|
✗ |
res.tail(6) = data.lambda_c; |
| 211 |
|
|
|
| 212 |
|
✗ |
return res; |
| 213 |
|
|
} |
| 214 |
|
|
|
| 215 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_FD_with_contact_varying_gamma) |
| 216 |
|
|
{ |
| 217 |
|
✗ |
pinocchio::Model model; |
| 218 |
|
✗ |
pinocchio::buildModels::humanoidRandom(model, true); |
| 219 |
|
✗ |
pinocchio::Data data(model), data_check(model); |
| 220 |
|
|
|
| 221 |
|
✗ |
VectorXd q = VectorXd::Ones(model.nq); |
| 222 |
|
✗ |
normalize(model, q); |
| 223 |
|
|
|
| 224 |
|
✗ |
VectorXd v = VectorXd::Random(model.nv); |
| 225 |
|
✗ |
VectorXd tau = VectorXd::Random(model.nv); |
| 226 |
|
|
|
| 227 |
|
✗ |
const std::string RF = "rleg6_joint"; |
| 228 |
|
✗ |
const Model::JointIndex RF_id = model.getJointId(RF); |
| 229 |
|
|
|
| 230 |
|
✗ |
Data::Matrix6x J_RF(6, model.nv); |
| 231 |
|
✗ |
J_RF.setZero(); |
| 232 |
|
✗ |
computeJointJacobians(model, data, q); |
| 233 |
|
✗ |
getJointJacobian(model, data, RF_id, LOCAL, J_RF); |
| 234 |
|
✗ |
Motion::Vector6 gamma_RF; |
| 235 |
|
✗ |
gamma_RF.setZero(); |
| 236 |
|
|
|
| 237 |
|
✗ |
VectorXd x_ref = constraintDynamics(model, data, q, v, tau, RF_id); |
| 238 |
|
✗ |
VectorXd ddq_ref = x_ref.head(model.nv); |
| 239 |
|
✗ |
Force::Vector6 contact_force_ref = x_ref.tail(6); |
| 240 |
|
|
|
| 241 |
|
✗ |
container::aligned_vector<Force> fext((size_t)model.njoints, Force::Zero()); |
| 242 |
|
✗ |
fext[RF_id] = ForceRef<Force::Vector6>(contact_force_ref); |
| 243 |
|
|
|
| 244 |
|
|
// check call to RNEA |
| 245 |
|
✗ |
rnea(model, data_check, q, v, ddq_ref, fext); |
| 246 |
|
|
|
| 247 |
|
✗ |
BOOST_CHECK(data_check.tau.isApprox(tau)); |
| 248 |
|
✗ |
forwardKinematics(model, data_check, q, v, ddq_ref); |
| 249 |
|
✗ |
BOOST_CHECK(data_check.a[RF_id].toVector().isZero()); |
| 250 |
|
|
|
| 251 |
|
✗ |
Data data_fd(model); |
| 252 |
|
✗ |
VectorXd q_plus(model.nq); |
| 253 |
|
✗ |
VectorXd v_eps(model.nv); |
| 254 |
|
✗ |
v_eps.setZero(); |
| 255 |
|
✗ |
VectorXd v_plus(v); |
| 256 |
|
✗ |
VectorXd tau_plus(tau); |
| 257 |
|
✗ |
VectorXd x_plus(model.nv + 6); |
| 258 |
|
✗ |
const double eps = 1e-8; |
| 259 |
|
|
|
| 260 |
|
|
// check: dddq_dtau and dlambda_dtau |
| 261 |
|
✗ |
MatrixXd dddq_dtau(model.nv, model.nv); |
| 262 |
|
✗ |
Data::Matrix6x dlambda_dtau(6, model.nv); |
| 263 |
|
|
|
| 264 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 265 |
|
|
{ |
| 266 |
|
✗ |
tau_plus[k] += eps; |
| 267 |
|
✗ |
x_plus = constraintDynamics(model, data, q, v, tau_plus, RF_id); |
| 268 |
|
|
|
| 269 |
|
✗ |
const Data::TangentVectorType ddq_plus = x_plus.head(model.nv); |
| 270 |
|
✗ |
Force::Vector6 contact_force_plus = x_plus.tail(6); |
| 271 |
|
|
|
| 272 |
|
✗ |
dddq_dtau.col(k) = (ddq_plus - ddq_ref) / eps; |
| 273 |
|
✗ |
dlambda_dtau.col(k) = (contact_force_plus - contact_force_ref) / eps; |
| 274 |
|
|
|
| 275 |
|
✗ |
tau_plus[k] -= eps; |
| 276 |
|
|
} |
| 277 |
|
|
|
| 278 |
|
✗ |
MatrixXd A(model.nv + 6, model.nv + 6); |
| 279 |
|
✗ |
data.M.transpose().triangularView<Eigen::Upper>() = data.M.triangularView<Eigen::Upper>(); |
| 280 |
|
✗ |
A.topLeftCorner(model.nv, model.nv) = data.M; |
| 281 |
|
✗ |
A.bottomLeftCorner(6, model.nv) = J_RF; |
| 282 |
|
✗ |
A.topRightCorner(model.nv, 6) = J_RF.transpose(); |
| 283 |
|
✗ |
A.bottomRightCorner(6, 6).setZero(); |
| 284 |
|
|
|
| 285 |
|
✗ |
MatrixXd Ainv = A.inverse(); |
| 286 |
|
✗ |
BOOST_CHECK(Ainv.topRows(model.nv).leftCols(model.nv).isApprox(dddq_dtau, std::sqrt(eps))); |
| 287 |
|
✗ |
BOOST_CHECK(Ainv.bottomRows(6).leftCols(model.nv).isApprox(-dlambda_dtau, std::sqrt(eps))); |
| 288 |
|
|
|
| 289 |
|
|
// check: dddq_dv and dlambda_dv |
| 290 |
|
✗ |
MatrixXd dddq_dv(model.nv, model.nv); |
| 291 |
|
✗ |
Data::Matrix6x dlambda_dv(6, model.nv); |
| 292 |
|
|
|
| 293 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 294 |
|
|
{ |
| 295 |
|
✗ |
v_plus[k] += eps; |
| 296 |
|
✗ |
x_plus = constraintDynamics(model, data, q, v_plus, tau, RF_id); |
| 297 |
|
|
|
| 298 |
|
✗ |
const Data::TangentVectorType ddq_plus = x_plus.head(model.nv); |
| 299 |
|
✗ |
Force::Vector6 contact_force_plus = x_plus.tail(6); |
| 300 |
|
|
|
| 301 |
|
✗ |
dddq_dv.col(k) = (ddq_plus - ddq_ref) / eps; |
| 302 |
|
✗ |
dlambda_dv.col(k) = (contact_force_plus - contact_force_ref) / eps; |
| 303 |
|
|
|
| 304 |
|
✗ |
v_plus[k] -= eps; |
| 305 |
|
|
} |
| 306 |
|
|
|
| 307 |
|
✗ |
computeRNEADerivatives(model, data_check, q, v, VectorXd::Zero(model.nv)); |
| 308 |
|
✗ |
Data::Matrix6x v_partial_dq(6, model.nv), a_partial_dq(6, model.nv), a_partial_dv(6, model.nv), |
| 309 |
|
✗ |
a_partial_da(6, model.nv); |
| 310 |
|
✗ |
v_partial_dq.setZero(); |
| 311 |
|
✗ |
a_partial_dq.setZero(); |
| 312 |
|
✗ |
a_partial_dv.setZero(); |
| 313 |
|
✗ |
a_partial_da.setZero(); |
| 314 |
|
✗ |
Data data_kin(model); |
| 315 |
|
✗ |
computeForwardKinematicsDerivatives(model, data_kin, q, v, VectorXd::Zero(model.nv)); |
| 316 |
|
✗ |
getJointAccelerationDerivatives( |
| 317 |
|
|
model, data_kin, RF_id, LOCAL, v_partial_dq, a_partial_dq, a_partial_dv, a_partial_da); |
| 318 |
|
|
|
| 319 |
|
✗ |
MatrixXd dddq_dv_anal = -Ainv.topRows(model.nv).leftCols(model.nv) * data_check.dtau_dv; |
| 320 |
|
✗ |
dddq_dv_anal -= Ainv.topRows(model.nv).rightCols(6) * a_partial_dv; |
| 321 |
|
✗ |
MatrixXd dlambda_dv_anal = -Ainv.bottomRows(6).leftCols(model.nv) * data_check.dtau_dv; |
| 322 |
|
✗ |
dlambda_dv_anal -= Ainv.bottomRows(6).rightCols(6) * a_partial_dv; |
| 323 |
|
|
|
| 324 |
|
✗ |
BOOST_CHECK(dddq_dv_anal.isApprox(dddq_dv, std::sqrt(eps))); |
| 325 |
|
✗ |
BOOST_CHECK(dlambda_dv_anal.isApprox(-dlambda_dv, std::sqrt(eps))); |
| 326 |
|
|
|
| 327 |
|
✗ |
MatrixXd dddq_dq(model.nv, model.nv); |
| 328 |
|
✗ |
Data::Matrix6x dlambda_dq(6, model.nv); |
| 329 |
|
|
|
| 330 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 331 |
|
|
{ |
| 332 |
|
✗ |
v_eps[k] = eps; |
| 333 |
|
✗ |
q_plus = integrate(model, q, v_eps); |
| 334 |
|
|
|
| 335 |
|
✗ |
x_plus = constraintDynamics(model, data, q_plus, v, tau, RF_id); |
| 336 |
|
|
|
| 337 |
|
✗ |
const Data::TangentVectorType ddq_plus = x_plus.head(model.nv); |
| 338 |
|
✗ |
Force::Vector6 contact_force_plus = x_plus.tail(6); |
| 339 |
|
|
|
| 340 |
|
✗ |
dddq_dq.col(k) = (ddq_plus - ddq_ref) / eps; |
| 341 |
|
✗ |
dlambda_dq.col(k) = (contact_force_plus - contact_force_ref) / eps; |
| 342 |
|
|
|
| 343 |
|
✗ |
v_eps[k] = 0.; |
| 344 |
|
|
} |
| 345 |
|
|
|
| 346 |
|
✗ |
computeRNEADerivatives(model, data_check, q, v, ddq_ref, fext); |
| 347 |
|
✗ |
v_partial_dq.setZero(); |
| 348 |
|
✗ |
a_partial_dq.setZero(); |
| 349 |
|
✗ |
a_partial_dv.setZero(); |
| 350 |
|
✗ |
a_partial_da.setZero(); |
| 351 |
|
✗ |
computeForwardKinematicsDerivatives(model, data_kin, q, v, ddq_ref); |
| 352 |
|
✗ |
getJointAccelerationDerivatives( |
| 353 |
|
|
model, data_kin, RF_id, LOCAL, v_partial_dq, a_partial_dq, a_partial_dv, a_partial_da); |
| 354 |
|
|
|
| 355 |
|
✗ |
MatrixXd dddq_dq_anal = -Ainv.topRows(model.nv).leftCols(model.nv) * data_check.dtau_dq; |
| 356 |
|
✗ |
dddq_dq_anal -= Ainv.topRows(model.nv).rightCols(6) * a_partial_dq; |
| 357 |
|
|
|
| 358 |
|
✗ |
BOOST_CHECK(dddq_dq_anal.isApprox(dddq_dq, std::sqrt(eps))); |
| 359 |
|
|
|
| 360 |
|
✗ |
MatrixXd dlambda_dq_anal = Ainv.bottomRows(6).leftCols(model.nv) * data_check.dtau_dq; |
| 361 |
|
✗ |
dlambda_dq_anal += Ainv.bottomRows(6).rightCols(6) * a_partial_dq; |
| 362 |
|
|
|
| 363 |
|
✗ |
BOOST_CHECK(dlambda_dq_anal.isApprox(dlambda_dq, std::sqrt(eps))); |
| 364 |
|
|
} |
| 365 |
|
|
|
| 366 |
|
|
BOOST_AUTO_TEST_SUITE_END() |
| 367 |
|
|
|