| Line |
Branch |
Exec |
Source |
| 1 |
|
|
// |
| 2 |
|
|
// Copyright (c) 2017-2020 CNRS INRIA |
| 3 |
|
|
// |
| 4 |
|
|
|
| 5 |
|
|
#include "pinocchio/multibody/model.hpp" |
| 6 |
|
|
#include "pinocchio/multibody/data.hpp" |
| 7 |
|
|
#include "pinocchio/algorithm/jacobian.hpp" |
| 8 |
|
|
#include "pinocchio/algorithm/joint-configuration.hpp" |
| 9 |
|
|
#include "pinocchio/algorithm/kinematics.hpp" |
| 10 |
|
|
#include "pinocchio/algorithm/kinematics-derivatives.hpp" |
| 11 |
|
|
#include "pinocchio/algorithm/rnea.hpp" |
| 12 |
|
|
#include "pinocchio/algorithm/rnea-derivatives.hpp" |
| 13 |
|
|
#include "pinocchio/algorithm/crba.hpp" |
| 14 |
|
|
#include "pinocchio/multibody/sample-models.hpp" |
| 15 |
|
|
|
| 16 |
|
|
#include <iostream> |
| 17 |
|
|
|
| 18 |
|
|
#include <boost/test/unit_test.hpp> |
| 19 |
|
|
#include <boost/utility/binary.hpp> |
| 20 |
|
|
|
| 21 |
|
|
BOOST_AUTO_TEST_SUITE(BOOST_TEST_MODULE) |
| 22 |
|
|
|
| 23 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_generalized_gravity_derivatives) |
| 24 |
|
|
{ |
| 25 |
|
|
using namespace Eigen; |
| 26 |
|
|
using namespace pinocchio; |
| 27 |
|
|
|
| 28 |
|
✗ |
Model model; |
| 29 |
|
✗ |
buildModels::humanoidRandom(model); |
| 30 |
|
|
|
| 31 |
|
✗ |
Data data(model), data_fd(model); |
| 32 |
|
|
|
| 33 |
|
✗ |
model.lowerPositionLimit.head<3>().fill(-1.); |
| 34 |
|
✗ |
model.upperPositionLimit.head<3>().fill(1.); |
| 35 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 36 |
|
✗ |
VectorXd v(VectorXd::Zero(model.nv)); |
| 37 |
|
✗ |
VectorXd a(VectorXd::Zero(model.nv)); |
| 38 |
|
|
|
| 39 |
|
|
// Check againt non-derivative algo |
| 40 |
|
✗ |
MatrixXd g_partial_dq(model.nv, model.nv); |
| 41 |
|
✗ |
g_partial_dq.setZero(); |
| 42 |
|
✗ |
computeGeneralizedGravityDerivatives(model, data, q, g_partial_dq); |
| 43 |
|
|
|
| 44 |
|
✗ |
VectorXd g0 = computeGeneralizedGravity(model, data_fd, q); |
| 45 |
|
✗ |
BOOST_CHECK(data.g.isApprox(g0)); |
| 46 |
|
|
|
| 47 |
|
✗ |
MatrixXd g_partial_dq_fd(model.nv, model.nv); |
| 48 |
|
✗ |
g_partial_dq_fd.setZero(); |
| 49 |
|
|
|
| 50 |
|
✗ |
VectorXd v_eps(Eigen::VectorXd::Zero(model.nv)); |
| 51 |
|
✗ |
VectorXd q_plus(model.nq); |
| 52 |
|
✗ |
VectorXd g_plus(model.nv); |
| 53 |
|
✗ |
const double alpha = 1e-8; |
| 54 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 55 |
|
|
{ |
| 56 |
|
✗ |
v_eps[k] += alpha; |
| 57 |
|
✗ |
q_plus = integrate(model, q, v_eps); |
| 58 |
|
✗ |
g_plus = computeGeneralizedGravity(model, data_fd, q_plus); |
| 59 |
|
|
|
| 60 |
|
✗ |
g_partial_dq_fd.col(k) = (g_plus - g0) / alpha; |
| 61 |
|
✗ |
v_eps[k] -= alpha; |
| 62 |
|
|
} |
| 63 |
|
|
|
| 64 |
|
✗ |
BOOST_CHECK(g_partial_dq.isApprox(g_partial_dq_fd, sqrt(alpha))); |
| 65 |
|
|
} |
| 66 |
|
|
|
| 67 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_generalized_gravity_derivatives_fext) |
| 68 |
|
|
{ |
| 69 |
|
|
using namespace Eigen; |
| 70 |
|
|
using namespace pinocchio; |
| 71 |
|
|
|
| 72 |
|
✗ |
Model model; |
| 73 |
|
✗ |
buildModels::humanoidRandom(model); |
| 74 |
|
|
|
| 75 |
|
✗ |
Data data(model), data_fd(model); |
| 76 |
|
|
|
| 77 |
|
✗ |
model.lowerPositionLimit.head<3>().fill(-1.); |
| 78 |
|
✗ |
model.upperPositionLimit.head<3>().fill(1.); |
| 79 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 80 |
|
|
|
| 81 |
|
|
typedef PINOCCHIO_ALIGNED_STD_VECTOR(Force) ForceVector; |
| 82 |
|
✗ |
ForceVector fext((size_t)model.njoints); |
| 83 |
|
✗ |
for (ForceVector::iterator it = fext.begin(); it != fext.end(); ++it) |
| 84 |
|
✗ |
(*it).setRandom(); |
| 85 |
|
|
|
| 86 |
|
|
// Check againt non-derivative algo |
| 87 |
|
✗ |
MatrixXd static_vec_partial_dq(model.nv, model.nv); |
| 88 |
|
✗ |
static_vec_partial_dq.setZero(); |
| 89 |
|
✗ |
computeStaticTorqueDerivatives(model, data, q, fext, static_vec_partial_dq); |
| 90 |
|
|
|
| 91 |
|
✗ |
VectorXd tau0 = computeStaticTorque(model, data_fd, q, fext); |
| 92 |
|
✗ |
BOOST_CHECK(data.tau.isApprox(tau0)); |
| 93 |
|
|
|
| 94 |
|
✗ |
MatrixXd static_vec_partial_dq_fd(model.nv, model.nv); |
| 95 |
|
|
|
| 96 |
|
✗ |
VectorXd v_eps(Eigen::VectorXd::Zero(model.nv)); |
| 97 |
|
✗ |
VectorXd q_plus(model.nq); |
| 98 |
|
✗ |
VectorXd tau_plus(model.nv); |
| 99 |
|
✗ |
const double alpha = 1e-8; |
| 100 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 101 |
|
|
{ |
| 102 |
|
✗ |
v_eps[k] += alpha; |
| 103 |
|
✗ |
q_plus = integrate(model, q, v_eps); |
| 104 |
|
✗ |
tau_plus = computeStaticTorque(model, data_fd, q_plus, fext); |
| 105 |
|
|
|
| 106 |
|
✗ |
static_vec_partial_dq_fd.col(k) = (tau_plus - tau0) / alpha; |
| 107 |
|
✗ |
v_eps[k] = 0.; |
| 108 |
|
|
} |
| 109 |
|
|
|
| 110 |
|
✗ |
BOOST_CHECK(static_vec_partial_dq.isApprox(static_vec_partial_dq_fd, sqrt(alpha))); |
| 111 |
|
|
} |
| 112 |
|
|
|
| 113 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_rnea_derivatives) |
| 114 |
|
|
{ |
| 115 |
|
|
using namespace Eigen; |
| 116 |
|
|
using namespace pinocchio; |
| 117 |
|
|
|
| 118 |
|
✗ |
Model model; |
| 119 |
|
✗ |
buildModels::humanoidRandom(model); |
| 120 |
|
✗ |
model.armature = VectorXd::Random(model.nv) + VectorXd::Ones(model.nv); |
| 121 |
|
|
|
| 122 |
|
✗ |
Data data(model), data_fd(model), data_ref(model); |
| 123 |
|
|
|
| 124 |
|
✗ |
model.lowerPositionLimit.head<3>().fill(-1.); |
| 125 |
|
✗ |
model.upperPositionLimit.head<3>().fill(1.); |
| 126 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 127 |
|
✗ |
VectorXd v(VectorXd::Random(model.nv)); |
| 128 |
|
✗ |
VectorXd a(VectorXd::Random(model.nv)); |
| 129 |
|
|
|
| 130 |
|
|
/// Check againt computeGeneralizedGravityDerivatives |
| 131 |
|
✗ |
MatrixXd rnea_partial_dq(model.nv, model.nv); |
| 132 |
|
✗ |
rnea_partial_dq.setZero(); |
| 133 |
|
✗ |
MatrixXd rnea_partial_dv(model.nv, model.nv); |
| 134 |
|
✗ |
rnea_partial_dv.setZero(); |
| 135 |
|
✗ |
MatrixXd rnea_partial_da(model.nv, model.nv); |
| 136 |
|
✗ |
rnea_partial_da.setZero(); |
| 137 |
|
✗ |
computeRNEADerivatives( |
| 138 |
|
✗ |
model, data, q, VectorXd::Zero(model.nv), VectorXd::Zero(model.nv), rnea_partial_dq, |
| 139 |
|
|
rnea_partial_dv, rnea_partial_da); |
| 140 |
|
✗ |
rnea(model, data_ref, q, VectorXd::Zero(model.nv), VectorXd::Zero(model.nv)); |
| 141 |
|
✗ |
for (Model::JointIndex k = 1; k < (Model::JointIndex)model.njoints; ++k) |
| 142 |
|
|
{ |
| 143 |
|
✗ |
BOOST_CHECK(data.of[k].isApprox(data.oMi[k].act(data_ref.f[k]))); |
| 144 |
|
|
} |
| 145 |
|
|
|
| 146 |
|
✗ |
MatrixXd g_partial_dq(model.nv, model.nv); |
| 147 |
|
✗ |
g_partial_dq.setZero(); |
| 148 |
|
✗ |
computeGeneralizedGravityDerivatives(model, data_ref, q, g_partial_dq); |
| 149 |
|
|
|
| 150 |
|
✗ |
BOOST_CHECK(data.dFdq.isApprox(data_ref.dFdq)); |
| 151 |
|
✗ |
BOOST_CHECK(rnea_partial_dq.isApprox(g_partial_dq)); |
| 152 |
|
✗ |
BOOST_CHECK(data.tau.isApprox(data_ref.g)); |
| 153 |
|
|
|
| 154 |
|
✗ |
VectorXd tau0 = rnea(model, data_fd, q, VectorXd::Zero(model.nv), VectorXd::Zero(model.nv)); |
| 155 |
|
✗ |
MatrixXd rnea_partial_dq_fd(model.nv, model.nv); |
| 156 |
|
✗ |
rnea_partial_dq_fd.setZero(); |
| 157 |
|
|
|
| 158 |
|
✗ |
VectorXd v_eps(VectorXd::Zero(model.nv)); |
| 159 |
|
✗ |
VectorXd q_plus(model.nq); |
| 160 |
|
✗ |
VectorXd tau_plus(model.nv); |
| 161 |
|
✗ |
const double alpha = 1e-8; |
| 162 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 163 |
|
|
{ |
| 164 |
|
✗ |
v_eps[k] += alpha; |
| 165 |
|
✗ |
q_plus = integrate(model, q, v_eps); |
| 166 |
|
✗ |
tau_plus = rnea(model, data_fd, q_plus, VectorXd::Zero(model.nv), VectorXd::Zero(model.nv)); |
| 167 |
|
|
|
| 168 |
|
✗ |
rnea_partial_dq_fd.col(k) = (tau_plus - tau0) / alpha; |
| 169 |
|
✗ |
v_eps[k] -= alpha; |
| 170 |
|
|
} |
| 171 |
|
✗ |
BOOST_CHECK(rnea_partial_dq.isApprox(rnea_partial_dq_fd, sqrt(alpha))); |
| 172 |
|
|
|
| 173 |
|
|
// Check with q and a non zero |
| 174 |
|
✗ |
tau0 = rnea(model, data_fd, q, 0 * v, a); |
| 175 |
|
✗ |
rnea_partial_dq_fd.setZero(); |
| 176 |
|
|
|
| 177 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 178 |
|
|
{ |
| 179 |
|
✗ |
v_eps[k] += alpha; |
| 180 |
|
✗ |
q_plus = integrate(model, q, v_eps); |
| 181 |
|
✗ |
tau_plus = rnea(model, data_fd, q_plus, VectorXd::Zero(model.nv), a); |
| 182 |
|
|
|
| 183 |
|
✗ |
rnea_partial_dq_fd.col(k) = (tau_plus - tau0) / alpha; |
| 184 |
|
✗ |
v_eps[k] -= alpha; |
| 185 |
|
|
} |
| 186 |
|
|
|
| 187 |
|
✗ |
rnea_partial_dq.setZero(); |
| 188 |
|
✗ |
computeRNEADerivatives( |
| 189 |
|
✗ |
model, data, q, VectorXd::Zero(model.nv), a, rnea_partial_dq, rnea_partial_dv, rnea_partial_da); |
| 190 |
|
✗ |
forwardKinematics(model, data_ref, q, VectorXd::Zero(model.nv), a); |
| 191 |
|
|
|
| 192 |
|
✗ |
for (Model::JointIndex k = 1; k < (Model::JointIndex)model.njoints; ++k) |
| 193 |
|
|
{ |
| 194 |
|
✗ |
BOOST_CHECK(data.a[k].isApprox(data_ref.a[k])); |
| 195 |
|
✗ |
BOOST_CHECK(data.v[k].isApprox(data_ref.v[k])); |
| 196 |
|
✗ |
BOOST_CHECK(data.oMi[k].isApprox(data_ref.oMi[k])); |
| 197 |
|
✗ |
BOOST_CHECK(data.oh[k].isApprox(Force::Zero())); |
| 198 |
|
|
} |
| 199 |
|
|
|
| 200 |
|
✗ |
BOOST_CHECK(data.tau.isApprox(tau0)); |
| 201 |
|
✗ |
BOOST_CHECK(rnea_partial_dq.isApprox(rnea_partial_dq_fd, sqrt(alpha))); |
| 202 |
|
|
|
| 203 |
|
|
// Check with q and v non zero |
| 204 |
|
✗ |
const Motion gravity(model.gravity); |
| 205 |
|
✗ |
model.gravity.setZero(); |
| 206 |
|
✗ |
tau0 = rnea(model, data_fd, q, v, VectorXd::Zero(model.nv)); |
| 207 |
|
✗ |
rnea_partial_dq_fd.setZero(); |
| 208 |
|
|
|
| 209 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 210 |
|
|
{ |
| 211 |
|
✗ |
v_eps[k] += alpha; |
| 212 |
|
✗ |
q_plus = integrate(model, q, v_eps); |
| 213 |
|
✗ |
tau_plus = rnea(model, data_fd, q_plus, v, VectorXd::Zero(model.nv)); |
| 214 |
|
|
|
| 215 |
|
✗ |
rnea_partial_dq_fd.col(k) = (tau_plus - tau0) / alpha; |
| 216 |
|
✗ |
v_eps[k] -= alpha; |
| 217 |
|
|
} |
| 218 |
|
|
|
| 219 |
|
✗ |
VectorXd v_plus(v); |
| 220 |
|
✗ |
MatrixXd rnea_partial_dv_fd(model.nv, model.nv); |
| 221 |
|
✗ |
rnea_partial_dv_fd.setZero(); |
| 222 |
|
|
|
| 223 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 224 |
|
|
{ |
| 225 |
|
✗ |
v_plus[k] += alpha; |
| 226 |
|
✗ |
tau_plus = rnea(model, data_fd, q, v_plus, VectorXd::Zero(model.nv)); |
| 227 |
|
|
|
| 228 |
|
✗ |
rnea_partial_dv_fd.col(k) = (tau_plus - tau0) / alpha; |
| 229 |
|
✗ |
v_plus[k] -= alpha; |
| 230 |
|
|
} |
| 231 |
|
|
|
| 232 |
|
✗ |
rnea_partial_dq.setZero(); |
| 233 |
|
✗ |
rnea_partial_dv.setZero(); |
| 234 |
|
✗ |
computeRNEADerivatives( |
| 235 |
|
✗ |
model, data, q, v, VectorXd::Zero(model.nv), rnea_partial_dq, rnea_partial_dv, rnea_partial_da); |
| 236 |
|
✗ |
forwardKinematics(model, data_ref, q, v, VectorXd::Zero(model.nv)); |
| 237 |
|
|
|
| 238 |
|
✗ |
for (Model::JointIndex k = 1; k < (Model::JointIndex)model.njoints; ++k) |
| 239 |
|
|
{ |
| 240 |
|
✗ |
BOOST_CHECK(data.a[k].isApprox(data_ref.a[k])); |
| 241 |
|
✗ |
BOOST_CHECK(data.v[k].isApprox(data_ref.v[k])); |
| 242 |
|
✗ |
BOOST_CHECK(data.oMi[k].isApprox(data_ref.oMi[k])); |
| 243 |
|
|
} |
| 244 |
|
|
|
| 245 |
|
✗ |
BOOST_CHECK(data.tau.isApprox(tau0)); |
| 246 |
|
✗ |
BOOST_CHECK(rnea_partial_dq.isApprox(rnea_partial_dq_fd, sqrt(alpha))); |
| 247 |
|
✗ |
BOOST_CHECK(rnea_partial_dv.isApprox(rnea_partial_dv_fd, sqrt(alpha))); |
| 248 |
|
|
|
| 249 |
|
|
// std::cout << "rnea_partial_dv:\n" << rnea_partial_dv.block<10,10>(0,0) << std::endl; |
| 250 |
|
|
// std::cout << "rnea_partial_dv ref:\n" << rnea_partial_dv_fd.block<10,10>(0,0) << std::endl; |
| 251 |
|
|
// std::cout << "rnea_partial_dv:\n" << rnea_partial_dv.topRows<10>() << std::endl; |
| 252 |
|
|
// std::cout << "rnea_partial_dv ref:\n" << rnea_partial_dv_fd.topRows<10>() << std::endl; |
| 253 |
|
|
// Check with q, v and a non zero |
| 254 |
|
✗ |
model.gravity = gravity; |
| 255 |
|
✗ |
v_plus = v; |
| 256 |
|
✗ |
tau0 = rnea(model, data_fd, q, v, a); |
| 257 |
|
✗ |
rnea_partial_dq_fd.setZero(); |
| 258 |
|
|
|
| 259 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 260 |
|
|
{ |
| 261 |
|
✗ |
v_eps[k] += alpha; |
| 262 |
|
✗ |
q_plus = integrate(model, q, v_eps); |
| 263 |
|
✗ |
tau_plus = rnea(model, data_fd, q_plus, v, a); |
| 264 |
|
|
|
| 265 |
|
✗ |
rnea_partial_dq_fd.col(k) = (tau_plus - tau0) / alpha; |
| 266 |
|
✗ |
v_eps[k] -= alpha; |
| 267 |
|
|
} |
| 268 |
|
|
|
| 269 |
|
✗ |
rnea_partial_dv_fd.setZero(); |
| 270 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 271 |
|
|
{ |
| 272 |
|
✗ |
v_plus[k] += alpha; |
| 273 |
|
✗ |
tau_plus = rnea(model, data_fd, q, v_plus, a); |
| 274 |
|
|
|
| 275 |
|
✗ |
rnea_partial_dv_fd.col(k) = (tau_plus - tau0) / alpha; |
| 276 |
|
✗ |
v_plus[k] -= alpha; |
| 277 |
|
|
} |
| 278 |
|
|
|
| 279 |
|
✗ |
rnea_partial_dq.setZero(); |
| 280 |
|
✗ |
rnea_partial_dv.setZero(); |
| 281 |
|
✗ |
computeRNEADerivatives(model, data, q, v, a, rnea_partial_dq, rnea_partial_dv, rnea_partial_da); |
| 282 |
|
✗ |
forwardKinematics(model, data_ref, q, v, a); |
| 283 |
|
|
|
| 284 |
|
✗ |
for (Model::JointIndex k = 1; k < (Model::JointIndex)model.njoints; ++k) |
| 285 |
|
|
{ |
| 286 |
|
✗ |
BOOST_CHECK(data.a[k].isApprox(data_ref.a[k])); |
| 287 |
|
✗ |
BOOST_CHECK(data.v[k].isApprox(data_ref.v[k])); |
| 288 |
|
✗ |
BOOST_CHECK(data.oMi[k].isApprox(data_ref.oMi[k])); |
| 289 |
|
|
} |
| 290 |
|
|
|
| 291 |
|
✗ |
computeJointJacobiansTimeVariation(model, data_ref, q, v); |
| 292 |
|
✗ |
BOOST_CHECK(data.dJ.isApprox(data_ref.dJ)); |
| 293 |
|
✗ |
crba(model, data_ref, q, Convention::WORLD); |
| 294 |
|
✗ |
data_ref.M.triangularView<Eigen::StrictlyLower>() = |
| 295 |
|
✗ |
data_ref.M.transpose().triangularView<Eigen::StrictlyLower>(); |
| 296 |
|
|
|
| 297 |
|
✗ |
rnea_partial_da.triangularView<Eigen::StrictlyLower>() = |
| 298 |
|
✗ |
rnea_partial_da.transpose().triangularView<Eigen::StrictlyLower>(); |
| 299 |
|
✗ |
BOOST_CHECK(rnea_partial_da.isApprox(data_ref.M)); |
| 300 |
|
|
|
| 301 |
|
✗ |
BOOST_CHECK(data.tau.isApprox(tau0)); |
| 302 |
|
✗ |
BOOST_CHECK(rnea_partial_dq.isApprox(rnea_partial_dq_fd, sqrt(alpha))); |
| 303 |
|
✗ |
BOOST_CHECK(rnea_partial_dv.isApprox(rnea_partial_dv_fd, sqrt(alpha))); |
| 304 |
|
|
|
| 305 |
|
✗ |
Data data2(model); |
| 306 |
|
✗ |
computeRNEADerivatives(model, data2, q, v, a); |
| 307 |
|
✗ |
data2.M.triangularView<Eigen::StrictlyLower>() = |
| 308 |
|
✗ |
data2.M.transpose().triangularView<Eigen::StrictlyLower>(); |
| 309 |
|
|
|
| 310 |
|
✗ |
BOOST_CHECK(rnea_partial_dq.isApprox(data2.dtau_dq)); |
| 311 |
|
✗ |
BOOST_CHECK(rnea_partial_dv.isApprox(data2.dtau_dv)); |
| 312 |
|
✗ |
BOOST_CHECK(rnea_partial_da.isApprox(data2.M)); |
| 313 |
|
|
} |
| 314 |
|
|
|
| 315 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_rnea_derivatives_fext) |
| 316 |
|
|
{ |
| 317 |
|
|
using namespace Eigen; |
| 318 |
|
|
using namespace pinocchio; |
| 319 |
|
|
|
| 320 |
|
✗ |
Model model; |
| 321 |
|
✗ |
buildModels::humanoidRandom(model); |
| 322 |
|
|
typedef Model::Force Force; |
| 323 |
|
|
|
| 324 |
|
✗ |
Data data(model), data_fd(model), data_ref(model); |
| 325 |
|
|
|
| 326 |
|
✗ |
model.lowerPositionLimit.head<3>().fill(-1.); |
| 327 |
|
✗ |
model.upperPositionLimit.head<3>().fill(1.); |
| 328 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 329 |
|
✗ |
VectorXd v(VectorXd::Random(model.nv)); |
| 330 |
|
✗ |
VectorXd a(VectorXd::Random(model.nv)); |
| 331 |
|
|
|
| 332 |
|
|
typedef PINOCCHIO_ALIGNED_STD_VECTOR(Force) ForceVector; |
| 333 |
|
✗ |
ForceVector fext((size_t)model.njoints); |
| 334 |
|
✗ |
for (ForceVector::iterator it = fext.begin(); it != fext.end(); ++it) |
| 335 |
|
✗ |
(*it).setRandom(); |
| 336 |
|
|
|
| 337 |
|
|
/// Check againt computeGeneralizedGravityDerivatives |
| 338 |
|
✗ |
MatrixXd rnea_partial_dq(model.nv, model.nv); |
| 339 |
|
✗ |
rnea_partial_dq.setZero(); |
| 340 |
|
✗ |
MatrixXd rnea_partial_dv(model.nv, model.nv); |
| 341 |
|
✗ |
rnea_partial_dv.setZero(); |
| 342 |
|
✗ |
MatrixXd rnea_partial_da(model.nv, model.nv); |
| 343 |
|
✗ |
rnea_partial_da.setZero(); |
| 344 |
|
|
|
| 345 |
|
✗ |
computeRNEADerivatives( |
| 346 |
|
|
model, data, q, v, a, fext, rnea_partial_dq, rnea_partial_dv, rnea_partial_da); |
| 347 |
|
✗ |
rnea(model, data_ref, q, v, a, fext); |
| 348 |
|
|
|
| 349 |
|
✗ |
BOOST_CHECK(data.tau.isApprox(data_ref.tau)); |
| 350 |
|
|
|
| 351 |
|
✗ |
computeRNEADerivatives(model, data_ref, q, v, a); |
| 352 |
|
✗ |
BOOST_CHECK(rnea_partial_dv.isApprox(data_ref.dtau_dv)); |
| 353 |
|
✗ |
BOOST_CHECK(rnea_partial_da.isApprox(data_ref.M)); |
| 354 |
|
|
|
| 355 |
|
✗ |
MatrixXd rnea_partial_dq_fd(model.nv, model.nv); |
| 356 |
|
✗ |
rnea_partial_dq_fd.setZero(); |
| 357 |
|
✗ |
MatrixXd rnea_partial_dv_fd(model.nv, model.nv); |
| 358 |
|
✗ |
rnea_partial_dv_fd.setZero(); |
| 359 |
|
✗ |
MatrixXd rnea_partial_da_fd(model.nv, model.nv); |
| 360 |
|
✗ |
rnea_partial_da_fd.setZero(); |
| 361 |
|
|
|
| 362 |
|
✗ |
VectorXd v_eps(VectorXd::Zero(model.nv)); |
| 363 |
|
✗ |
VectorXd q_plus(model.nq); |
| 364 |
|
✗ |
VectorXd tau_plus(model.nv); |
| 365 |
|
✗ |
const double eps = 1e-8; |
| 366 |
|
|
|
| 367 |
|
✗ |
const VectorXd tau_ref = rnea(model, data_ref, q, v, a, fext); |
| 368 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 369 |
|
|
{ |
| 370 |
|
✗ |
v_eps[k] = eps; |
| 371 |
|
✗ |
q_plus = integrate(model, q, v_eps); |
| 372 |
|
✗ |
tau_plus = rnea(model, data_fd, q_plus, v, a, fext); |
| 373 |
|
|
|
| 374 |
|
✗ |
rnea_partial_dq_fd.col(k) = (tau_plus - tau_ref) / eps; |
| 375 |
|
|
|
| 376 |
|
✗ |
v_eps[k] = 0.; |
| 377 |
|
|
} |
| 378 |
|
✗ |
BOOST_CHECK(rnea_partial_dq.isApprox(rnea_partial_dq_fd, sqrt(eps))); |
| 379 |
|
|
|
| 380 |
|
✗ |
VectorXd v_plus(v); |
| 381 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 382 |
|
|
{ |
| 383 |
|
✗ |
v_plus[k] += eps; |
| 384 |
|
|
|
| 385 |
|
✗ |
tau_plus = rnea(model, data_fd, q, v_plus, a, fext); |
| 386 |
|
|
|
| 387 |
|
✗ |
rnea_partial_dv_fd.col(k) = (tau_plus - tau_ref) / eps; |
| 388 |
|
|
|
| 389 |
|
✗ |
v_plus[k] -= eps; |
| 390 |
|
|
} |
| 391 |
|
✗ |
BOOST_CHECK(rnea_partial_dv.isApprox(rnea_partial_dv_fd, sqrt(eps))); |
| 392 |
|
|
|
| 393 |
|
✗ |
VectorXd a_plus(a); |
| 394 |
|
✗ |
for (int k = 0; k < model.nv; ++k) |
| 395 |
|
|
{ |
| 396 |
|
✗ |
a_plus[k] += eps; |
| 397 |
|
|
|
| 398 |
|
✗ |
tau_plus = rnea(model, data_fd, q, v, a_plus, fext); |
| 399 |
|
|
|
| 400 |
|
✗ |
rnea_partial_da_fd.col(k) = (tau_plus - tau_ref) / eps; |
| 401 |
|
|
|
| 402 |
|
✗ |
a_plus[k] -= eps; |
| 403 |
|
|
} |
| 404 |
|
|
|
| 405 |
|
✗ |
rnea_partial_da.triangularView<Eigen::Lower>() = |
| 406 |
|
✗ |
rnea_partial_da.transpose().triangularView<Eigen::Lower>(); |
| 407 |
|
✗ |
BOOST_CHECK(rnea_partial_da.isApprox(rnea_partial_da_fd, sqrt(eps))); |
| 408 |
|
|
|
| 409 |
|
|
// test the shortcut |
| 410 |
|
✗ |
Data data_shortcut(model); |
| 411 |
|
✗ |
computeRNEADerivatives(model, data_shortcut, q, v, a, fext); |
| 412 |
|
✗ |
BOOST_CHECK(data_shortcut.dtau_dq.isApprox(rnea_partial_dq)); |
| 413 |
|
✗ |
BOOST_CHECK(data_shortcut.dtau_dv.isApprox(rnea_partial_dv)); |
| 414 |
|
✗ |
data_shortcut.M.triangularView<Eigen::Lower>() = |
| 415 |
|
✗ |
data_shortcut.M.transpose().triangularView<Eigen::Lower>(); |
| 416 |
|
✗ |
BOOST_CHECK(data_shortcut.M.isApprox(rnea_partial_da)); |
| 417 |
|
|
} |
| 418 |
|
|
|
| 419 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_rnea_derivatives_vs_kinematics_derivatives) |
| 420 |
|
|
{ |
| 421 |
|
|
using namespace Eigen; |
| 422 |
|
|
using namespace pinocchio; |
| 423 |
|
|
|
| 424 |
|
✗ |
Model model; |
| 425 |
|
✗ |
buildModels::humanoidRandom(model); |
| 426 |
|
|
|
| 427 |
|
✗ |
Data data(model), data_ref(model); |
| 428 |
|
|
|
| 429 |
|
✗ |
model.lowerPositionLimit.head<3>().fill(-1.); |
| 430 |
|
✗ |
model.upperPositionLimit.head<3>().fill(1.); |
| 431 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 432 |
|
✗ |
VectorXd v(VectorXd::Random(model.nv)); |
| 433 |
|
✗ |
VectorXd a(VectorXd::Random(model.nv)); |
| 434 |
|
|
|
| 435 |
|
|
/// Check againt computeGeneralizedGravityDerivatives |
| 436 |
|
✗ |
MatrixXd rnea_partial_dq(model.nv, model.nv); |
| 437 |
|
✗ |
rnea_partial_dq.setZero(); |
| 438 |
|
✗ |
MatrixXd rnea_partial_dv(model.nv, model.nv); |
| 439 |
|
✗ |
rnea_partial_dv.setZero(); |
| 440 |
|
✗ |
MatrixXd rnea_partial_da(model.nv, model.nv); |
| 441 |
|
✗ |
rnea_partial_da.setZero(); |
| 442 |
|
|
|
| 443 |
|
✗ |
computeRNEADerivatives(model, data, q, v, a, rnea_partial_dq, rnea_partial_dv, rnea_partial_da); |
| 444 |
|
✗ |
computeForwardKinematicsDerivatives(model, data_ref, q, v, a); |
| 445 |
|
|
|
| 446 |
|
✗ |
BOOST_CHECK(data.J.isApprox(data_ref.J)); |
| 447 |
|
✗ |
BOOST_CHECK(data.dJ.isApprox(data_ref.dJ)); |
| 448 |
|
|
|
| 449 |
|
✗ |
for (size_t k = 1; k < (size_t)model.njoints; ++k) |
| 450 |
|
|
{ |
| 451 |
|
✗ |
BOOST_CHECK(data.oMi[k].isApprox(data_ref.oMi[k])); |
| 452 |
|
✗ |
BOOST_CHECK(data.ov[k].isApprox(data_ref.ov[k])); |
| 453 |
|
✗ |
BOOST_CHECK(data.oa[k].isApprox(data_ref.oa[k])); |
| 454 |
|
|
} |
| 455 |
|
|
} |
| 456 |
|
|
|
| 457 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_multiple_calls) |
| 458 |
|
|
{ |
| 459 |
|
|
using namespace Eigen; |
| 460 |
|
|
using namespace pinocchio; |
| 461 |
|
|
|
| 462 |
|
✗ |
Model model; |
| 463 |
|
✗ |
buildModels::humanoidRandom(model); |
| 464 |
|
|
|
| 465 |
|
✗ |
Data data1(model), data2(model); |
| 466 |
|
|
|
| 467 |
|
✗ |
model.lowerPositionLimit.head<3>().fill(-1.); |
| 468 |
|
✗ |
model.upperPositionLimit.head<3>().fill(1.); |
| 469 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 470 |
|
✗ |
VectorXd v(VectorXd::Random(model.nv)); |
| 471 |
|
✗ |
VectorXd a(VectorXd::Random(model.nv)); |
| 472 |
|
|
|
| 473 |
|
✗ |
computeRNEADerivatives(model, data1, q, v, a); |
| 474 |
|
✗ |
data2 = data1; |
| 475 |
|
|
|
| 476 |
|
✗ |
for (int k = 0; k < 20; ++k) |
| 477 |
|
|
{ |
| 478 |
|
✗ |
computeRNEADerivatives(model, data1, q, v, a); |
| 479 |
|
|
} |
| 480 |
|
|
|
| 481 |
|
✗ |
BOOST_CHECK(data1.J.isApprox(data2.J)); |
| 482 |
|
✗ |
BOOST_CHECK(data1.dJ.isApprox(data2.dJ)); |
| 483 |
|
✗ |
BOOST_CHECK(data1.dVdq.isApprox(data2.dVdq)); |
| 484 |
|
✗ |
BOOST_CHECK(data1.dAdq.isApprox(data2.dAdq)); |
| 485 |
|
✗ |
BOOST_CHECK(data1.dAdv.isApprox(data2.dAdv)); |
| 486 |
|
|
|
| 487 |
|
✗ |
BOOST_CHECK(data1.dFdq.isApprox(data2.dFdq)); |
| 488 |
|
✗ |
BOOST_CHECK(data1.dFdv.isApprox(data2.dFdv)); |
| 489 |
|
✗ |
BOOST_CHECK(data1.dFda.isApprox(data2.dFda)); |
| 490 |
|
|
|
| 491 |
|
✗ |
BOOST_CHECK(data1.dtau_dq.isApprox(data2.dtau_dq)); |
| 492 |
|
✗ |
BOOST_CHECK(data1.dtau_dv.isApprox(data2.dtau_dv)); |
| 493 |
|
✗ |
BOOST_CHECK(data1.M.isApprox(data2.M)); |
| 494 |
|
|
} |
| 495 |
|
|
|
| 496 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_get_coriolis) |
| 497 |
|
|
{ |
| 498 |
|
|
using namespace Eigen; |
| 499 |
|
|
using namespace pinocchio; |
| 500 |
|
|
|
| 501 |
|
✗ |
Model model; |
| 502 |
|
✗ |
buildModels::humanoidRandom(model); |
| 503 |
|
|
|
| 504 |
|
✗ |
model.lowerPositionLimit.head<3>().fill(-1.); |
| 505 |
|
✗ |
model.upperPositionLimit.head<3>().fill(1.); |
| 506 |
|
|
|
| 507 |
|
✗ |
Data data_ref(model); |
| 508 |
|
✗ |
Data data(model); |
| 509 |
|
|
|
| 510 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 511 |
|
✗ |
VectorXd v = VectorXd::Random(model.nv); |
| 512 |
|
✗ |
VectorXd tau = VectorXd::Random(model.nv); |
| 513 |
|
|
|
| 514 |
|
✗ |
computeCoriolisMatrix(model, data_ref, q, v); |
| 515 |
|
|
|
| 516 |
|
✗ |
computeRNEADerivatives(model, data, q, v, tau); |
| 517 |
|
✗ |
getCoriolisMatrix(model, data); |
| 518 |
|
|
|
| 519 |
|
✗ |
BOOST_CHECK(data.J.isApprox(data_ref.J)); |
| 520 |
|
✗ |
BOOST_CHECK(data.dJ.isApprox(data_ref.dJ)); |
| 521 |
|
✗ |
BOOST_CHECK(data.Ag.isApprox(data_ref.Ag)); |
| 522 |
|
✗ |
for (JointIndex k = 1; k < model.joints.size(); ++k) |
| 523 |
|
|
{ |
| 524 |
|
✗ |
BOOST_CHECK(data.B[k].isApprox(data_ref.B[k])); |
| 525 |
|
✗ |
BOOST_CHECK(data.oYcrb[k].isApprox(data_ref.oYcrb[k])); |
| 526 |
|
|
} |
| 527 |
|
|
|
| 528 |
|
✗ |
BOOST_CHECK(data.C.isApprox(data_ref.C)); |
| 529 |
|
|
} |
| 530 |
|
|
|
| 531 |
|
|
BOOST_AUTO_TEST_SUITE_END() |
| 532 |
|
|
|