| Line |
Branch |
Exec |
Source |
| 1 |
|
|
// |
| 2 |
|
|
// Copyright (c) 2018-2024 CNRS INRIA |
| 3 |
|
|
// |
| 4 |
|
|
|
| 5 |
|
|
#include "pinocchio/spatial/fwd.hpp" |
| 6 |
|
|
#include "pinocchio/spatial/explog.hpp" |
| 7 |
|
|
#include "pinocchio/algorithm/regressor.hpp" |
| 8 |
|
|
#include "pinocchio/algorithm/rnea.hpp" |
| 9 |
|
|
#include "pinocchio/algorithm/frames.hpp" |
| 10 |
|
|
#include "pinocchio/algorithm/joint-configuration.hpp" |
| 11 |
|
|
#include "pinocchio/algorithm/center-of-mass.hpp" |
| 12 |
|
|
#include "pinocchio/multibody/sample-models.hpp" |
| 13 |
|
|
#include "pinocchio/algorithm/compute-all-terms.hpp" |
| 14 |
|
|
|
| 15 |
|
|
#include <iostream> |
| 16 |
|
|
|
| 17 |
|
|
#include <boost/test/unit_test.hpp> |
| 18 |
|
|
#include <boost/utility/binary.hpp> |
| 19 |
|
|
|
| 20 |
|
|
BOOST_AUTO_TEST_SUITE(BOOST_TEST_MODULE) |
| 21 |
|
|
|
| 22 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_kinematic_regressor_joint) |
| 23 |
|
|
{ |
| 24 |
|
|
using namespace Eigen; |
| 25 |
|
|
using namespace pinocchio; |
| 26 |
|
|
|
| 27 |
|
✗ |
pinocchio::Model model; |
| 28 |
|
✗ |
buildModels::humanoidRandom(model); |
| 29 |
|
|
|
| 30 |
|
✗ |
pinocchio::Data data(model); |
| 31 |
|
✗ |
pinocchio::Data data_ref(model); |
| 32 |
|
|
|
| 33 |
|
✗ |
model.lowerPositionLimit.head<7>().fill(-1.); |
| 34 |
|
✗ |
model.upperPositionLimit.head<7>().fill(1.); |
| 35 |
|
|
|
| 36 |
|
|
// const std::string joint_name = "larm5_joint"; |
| 37 |
|
|
// const JointIndex joint_id = model.getJointId(joint_name); |
| 38 |
|
|
|
| 39 |
|
✗ |
const VectorXd q = randomConfiguration(model); |
| 40 |
|
|
|
| 41 |
|
✗ |
forwardKinematics(model, data, q); |
| 42 |
|
|
|
| 43 |
|
✗ |
const double eps = 1e-8; |
| 44 |
|
✗ |
for (JointIndex joint_id = 1; joint_id < (JointIndex)model.njoints; ++joint_id) |
| 45 |
|
|
{ |
| 46 |
|
✗ |
Data::Matrix6x kinematic_regressor_L(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 47 |
|
✗ |
Data::Matrix6x kinematic_regressor_LWA(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 48 |
|
✗ |
Data::Matrix6x kinematic_regressor_W(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 49 |
|
|
|
| 50 |
|
✗ |
Data::Matrix6x kinematic_regressor_L_fd(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 51 |
|
✗ |
Data::Matrix6x kinematic_regressor_LWA_fd(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 52 |
|
✗ |
Data::Matrix6x kinematic_regressor_W_fd(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 53 |
|
|
|
| 54 |
|
✗ |
computeJointKinematicRegressor(model, data, joint_id, LOCAL, kinematic_regressor_L); |
| 55 |
|
✗ |
computeJointKinematicRegressor( |
| 56 |
|
|
model, data, joint_id, LOCAL_WORLD_ALIGNED, kinematic_regressor_LWA); |
| 57 |
|
✗ |
computeJointKinematicRegressor(model, data, joint_id, WORLD, kinematic_regressor_W); |
| 58 |
|
|
|
| 59 |
|
✗ |
Model model_plus = model; |
| 60 |
|
✗ |
Data data_plus(model_plus); |
| 61 |
|
✗ |
const SE3 & oMi = data.oMi[joint_id]; |
| 62 |
|
✗ |
const SE3 Mi_LWA = SE3(oMi.rotation(), SE3::Vector3::Zero()); |
| 63 |
|
✗ |
const SE3 & oMi_plus = data_plus.oMi[joint_id]; |
| 64 |
|
✗ |
for (int i = 1; i < model.njoints; ++i) |
| 65 |
|
|
{ |
| 66 |
|
✗ |
Motion::Vector6 v = Motion::Vector6::Zero(); |
| 67 |
|
✗ |
const SE3 & M_placement = model.jointPlacements[(JointIndex)i]; |
| 68 |
|
✗ |
SE3 & M_placement_plus = model_plus.jointPlacements[(JointIndex)i]; |
| 69 |
|
✗ |
for (Eigen::DenseIndex k = 0; k < 6; ++k) |
| 70 |
|
|
{ |
| 71 |
|
✗ |
v[k] = eps; |
| 72 |
|
✗ |
M_placement_plus = M_placement * exp6(Motion(v)); |
| 73 |
|
|
|
| 74 |
|
✗ |
forwardKinematics(model_plus, data_plus, q); |
| 75 |
|
|
|
| 76 |
|
✗ |
const Motion diff_L = log6(oMi.actInv(oMi_plus)); |
| 77 |
|
✗ |
kinematic_regressor_L_fd.middleCols<6>(6 * (i - 1)).col(k) = diff_L.toVector() / eps; |
| 78 |
|
✗ |
const Motion diff_LWA = Mi_LWA.act(diff_L); |
| 79 |
|
✗ |
kinematic_regressor_LWA_fd.middleCols<6>(6 * (i - 1)).col(k) = diff_LWA.toVector() / eps; |
| 80 |
|
✗ |
const Motion diff_W = oMi.act(diff_L); |
| 81 |
|
✗ |
kinematic_regressor_W_fd.middleCols<6>(6 * (i - 1)).col(k) = diff_W.toVector() / eps; |
| 82 |
|
✗ |
v[k] = 0.; |
| 83 |
|
|
} |
| 84 |
|
|
|
| 85 |
|
✗ |
M_placement_plus = M_placement; |
| 86 |
|
|
} |
| 87 |
|
|
|
| 88 |
|
✗ |
BOOST_CHECK(kinematic_regressor_L.isApprox(kinematic_regressor_L_fd, sqrt(eps))); |
| 89 |
|
✗ |
BOOST_CHECK(kinematic_regressor_LWA.isApprox(kinematic_regressor_LWA_fd, sqrt(eps))); |
| 90 |
|
✗ |
BOOST_CHECK(kinematic_regressor_W.isApprox(kinematic_regressor_W_fd, sqrt(eps))); |
| 91 |
|
|
} |
| 92 |
|
|
} |
| 93 |
|
|
|
| 94 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_kinematic_regressor_joint_placement) |
| 95 |
|
|
{ |
| 96 |
|
|
using namespace Eigen; |
| 97 |
|
|
using namespace pinocchio; |
| 98 |
|
|
|
| 99 |
|
✗ |
pinocchio::Model model; |
| 100 |
|
✗ |
buildModels::humanoidRandom(model); |
| 101 |
|
|
|
| 102 |
|
✗ |
pinocchio::Data data(model); |
| 103 |
|
✗ |
pinocchio::Data data_ref(model); |
| 104 |
|
|
|
| 105 |
|
✗ |
model.lowerPositionLimit.head<7>().fill(-1.); |
| 106 |
|
✗ |
model.upperPositionLimit.head<7>().fill(1.); |
| 107 |
|
|
|
| 108 |
|
✗ |
const VectorXd q = randomConfiguration(model); |
| 109 |
|
|
|
| 110 |
|
✗ |
forwardKinematics(model, data, q); |
| 111 |
|
✗ |
forwardKinematics(model, data_ref, q); |
| 112 |
|
|
|
| 113 |
|
✗ |
for (JointIndex joint_id = 1; joint_id < (JointIndex)model.njoints; ++joint_id) |
| 114 |
|
|
{ |
| 115 |
|
✗ |
Data::Matrix6x kinematic_regressor_L(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 116 |
|
✗ |
Data::Matrix6x kinematic_regressor_LWA(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 117 |
|
✗ |
Data::Matrix6x kinematic_regressor_W(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 118 |
|
|
|
| 119 |
|
✗ |
computeJointKinematicRegressor( |
| 120 |
|
✗ |
model, data, joint_id, LOCAL, SE3::Identity(), kinematic_regressor_L); |
| 121 |
|
✗ |
computeJointKinematicRegressor( |
| 122 |
|
✗ |
model, data, joint_id, LOCAL_WORLD_ALIGNED, SE3::Identity(), kinematic_regressor_LWA); |
| 123 |
|
✗ |
computeJointKinematicRegressor( |
| 124 |
|
✗ |
model, data, joint_id, WORLD, SE3::Identity(), kinematic_regressor_W); |
| 125 |
|
|
|
| 126 |
|
✗ |
Data::Matrix6x kinematic_regressor_L_ref(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 127 |
|
✗ |
Data::Matrix6x kinematic_regressor_LWA_ref(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 128 |
|
✗ |
Data::Matrix6x kinematic_regressor_W_ref(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 129 |
|
|
|
| 130 |
|
✗ |
computeJointKinematicRegressor(model, data_ref, joint_id, LOCAL, kinematic_regressor_L_ref); |
| 131 |
|
✗ |
computeJointKinematicRegressor( |
| 132 |
|
|
model, data_ref, joint_id, LOCAL_WORLD_ALIGNED, kinematic_regressor_LWA_ref); |
| 133 |
|
✗ |
computeJointKinematicRegressor(model, data_ref, joint_id, WORLD, kinematic_regressor_W_ref); |
| 134 |
|
|
|
| 135 |
|
✗ |
BOOST_CHECK(kinematic_regressor_L.isApprox(kinematic_regressor_L_ref)); |
| 136 |
|
✗ |
BOOST_CHECK(kinematic_regressor_LWA.isApprox(kinematic_regressor_LWA_ref)); |
| 137 |
|
✗ |
BOOST_CHECK(kinematic_regressor_W.isApprox(kinematic_regressor_W_ref)); |
| 138 |
|
|
} |
| 139 |
|
|
} |
| 140 |
|
|
|
| 141 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_kinematic_regressor_frame) |
| 142 |
|
|
{ |
| 143 |
|
|
using namespace Eigen; |
| 144 |
|
|
using namespace pinocchio; |
| 145 |
|
|
|
| 146 |
|
✗ |
pinocchio::Model model; |
| 147 |
|
✗ |
buildModels::humanoidRandom(model); |
| 148 |
|
|
|
| 149 |
|
✗ |
model.lowerPositionLimit.head<7>().fill(-1.); |
| 150 |
|
✗ |
model.upperPositionLimit.head<7>().fill(1.); |
| 151 |
|
|
|
| 152 |
|
✗ |
const std::string joint_name = "larm5_joint"; |
| 153 |
|
✗ |
const JointIndex joint_id = model.getJointId(joint_name); |
| 154 |
|
✗ |
model.addBodyFrame("test_body", joint_id, SE3::Random(), -1); |
| 155 |
|
|
|
| 156 |
|
✗ |
pinocchio::Data data(model); |
| 157 |
|
✗ |
pinocchio::Data data_ref(model); |
| 158 |
|
|
|
| 159 |
|
✗ |
const VectorXd q = randomConfiguration(model); |
| 160 |
|
|
|
| 161 |
|
✗ |
forwardKinematics(model, data, q); |
| 162 |
|
✗ |
updateFramePlacements(model, data); |
| 163 |
|
✗ |
forwardKinematics(model, data_ref, q); |
| 164 |
|
|
|
| 165 |
|
✗ |
const double eps = 1e-8; |
| 166 |
|
✗ |
for (FrameIndex frame_id = 1; frame_id < (FrameIndex)model.nframes; ++frame_id) |
| 167 |
|
|
{ |
| 168 |
|
✗ |
const Frame & frame = model.frames[frame_id]; |
| 169 |
|
|
|
| 170 |
|
✗ |
Data::Matrix6x kinematic_regressor_L(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 171 |
|
✗ |
Data::Matrix6x kinematic_regressor_LWA(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 172 |
|
✗ |
Data::Matrix6x kinematic_regressor_W(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 173 |
|
|
|
| 174 |
|
✗ |
computeFrameKinematicRegressor(model, data, frame_id, LOCAL, kinematic_regressor_L); |
| 175 |
|
✗ |
computeFrameKinematicRegressor( |
| 176 |
|
|
model, data, frame_id, LOCAL_WORLD_ALIGNED, kinematic_regressor_LWA); |
| 177 |
|
✗ |
computeFrameKinematicRegressor(model, data, frame_id, WORLD, kinematic_regressor_W); |
| 178 |
|
|
|
| 179 |
|
✗ |
Data::Matrix6x kinematic_regressor_L_ref(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 180 |
|
✗ |
Data::Matrix6x kinematic_regressor_LWA_ref(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 181 |
|
✗ |
Data::Matrix6x kinematic_regressor_W_ref(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 182 |
|
|
|
| 183 |
|
✗ |
computeJointKinematicRegressor( |
| 184 |
|
✗ |
model, data_ref, frame.parentJoint, LOCAL, frame.placement, kinematic_regressor_L_ref); |
| 185 |
|
✗ |
computeJointKinematicRegressor( |
| 186 |
|
✗ |
model, data_ref, frame.parentJoint, LOCAL_WORLD_ALIGNED, frame.placement, |
| 187 |
|
|
kinematic_regressor_LWA_ref); |
| 188 |
|
✗ |
computeJointKinematicRegressor( |
| 189 |
|
✗ |
model, data_ref, frame.parentJoint, WORLD, frame.placement, kinematic_regressor_W_ref); |
| 190 |
|
|
|
| 191 |
|
✗ |
BOOST_CHECK(kinematic_regressor_L.isApprox(kinematic_regressor_L_ref)); |
| 192 |
|
✗ |
BOOST_CHECK(kinematic_regressor_LWA.isApprox(kinematic_regressor_LWA_ref)); |
| 193 |
|
✗ |
BOOST_CHECK(kinematic_regressor_W.isApprox(kinematic_regressor_W_ref)); |
| 194 |
|
|
|
| 195 |
|
✗ |
Data::Matrix6x kinematic_regressor_L_fd(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 196 |
|
✗ |
Data::Matrix6x kinematic_regressor_LWA_fd(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 197 |
|
✗ |
Data::Matrix6x kinematic_regressor_W_fd(Data::Matrix6x::Zero(6, 6 * (model.njoints - 1))); |
| 198 |
|
|
|
| 199 |
|
✗ |
Model model_plus = model; |
| 200 |
|
✗ |
Data data_plus(model_plus); |
| 201 |
|
✗ |
const SE3 & oMf = data.oMf[frame_id]; |
| 202 |
|
✗ |
const SE3 Mf_LWA = SE3(oMf.rotation(), SE3::Vector3::Zero()); |
| 203 |
|
✗ |
const SE3 & oMf_plus = data_plus.oMf[frame_id]; |
| 204 |
|
✗ |
for (int i = 1; i < model.njoints; ++i) |
| 205 |
|
|
{ |
| 206 |
|
✗ |
Motion::Vector6 v = Motion::Vector6::Zero(); |
| 207 |
|
✗ |
const SE3 & M_placement = model.jointPlacements[(JointIndex)i]; |
| 208 |
|
✗ |
SE3 & M_placement_plus = model_plus.jointPlacements[(JointIndex)i]; |
| 209 |
|
✗ |
for (Eigen::DenseIndex k = 0; k < 6; ++k) |
| 210 |
|
|
{ |
| 211 |
|
✗ |
v[k] = eps; |
| 212 |
|
✗ |
M_placement_plus = M_placement * exp6(Motion(v)); |
| 213 |
|
|
|
| 214 |
|
✗ |
forwardKinematics(model_plus, data_plus, q); |
| 215 |
|
✗ |
updateFramePlacements(model_plus, data_plus); |
| 216 |
|
|
|
| 217 |
|
✗ |
const Motion diff_L = log6(oMf.actInv(oMf_plus)); |
| 218 |
|
✗ |
kinematic_regressor_L_fd.middleCols<6>(6 * (i - 1)).col(k) = diff_L.toVector() / eps; |
| 219 |
|
✗ |
const Motion diff_LWA = Mf_LWA.act(diff_L); |
| 220 |
|
✗ |
kinematic_regressor_LWA_fd.middleCols<6>(6 * (i - 1)).col(k) = diff_LWA.toVector() / eps; |
| 221 |
|
✗ |
const Motion diff_W = oMf.act(diff_L); |
| 222 |
|
✗ |
kinematic_regressor_W_fd.middleCols<6>(6 * (i - 1)).col(k) = diff_W.toVector() / eps; |
| 223 |
|
✗ |
v[k] = 0.; |
| 224 |
|
|
} |
| 225 |
|
|
|
| 226 |
|
✗ |
M_placement_plus = M_placement; |
| 227 |
|
|
} |
| 228 |
|
|
|
| 229 |
|
✗ |
BOOST_CHECK(kinematic_regressor_L.isApprox(kinematic_regressor_L_fd, sqrt(eps))); |
| 230 |
|
✗ |
BOOST_CHECK(kinematic_regressor_LWA.isApprox(kinematic_regressor_LWA_fd, sqrt(eps))); |
| 231 |
|
✗ |
BOOST_CHECK(kinematic_regressor_W.isApprox(kinematic_regressor_W_fd, sqrt(eps))); |
| 232 |
|
|
} |
| 233 |
|
|
} |
| 234 |
|
|
|
| 235 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_static_regressor) |
| 236 |
|
|
{ |
| 237 |
|
|
using namespace Eigen; |
| 238 |
|
|
using namespace pinocchio; |
| 239 |
|
|
|
| 240 |
|
✗ |
pinocchio::Model model; |
| 241 |
|
✗ |
buildModels::humanoidRandom(model); |
| 242 |
|
|
|
| 243 |
|
✗ |
pinocchio::Data data(model); |
| 244 |
|
✗ |
pinocchio::Data data_ref(model); |
| 245 |
|
|
|
| 246 |
|
✗ |
model.lowerPositionLimit.head<7>().fill(-1.); |
| 247 |
|
✗ |
model.upperPositionLimit.head<7>().fill(1.); |
| 248 |
|
|
|
| 249 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 250 |
|
✗ |
computeStaticRegressor(model, data, q); |
| 251 |
|
|
|
| 252 |
|
✗ |
VectorXd phi(4 * (model.njoints - 1)); |
| 253 |
|
✗ |
for (int k = 1; k < model.njoints; ++k) |
| 254 |
|
|
{ |
| 255 |
|
✗ |
const Inertia & Y = model.inertias[(size_t)k]; |
| 256 |
|
✗ |
phi.segment<4>(4 * (k - 1)) << Y.mass(), Y.mass() * Y.lever(); |
| 257 |
|
|
} |
| 258 |
|
|
|
| 259 |
|
✗ |
Vector3d com = centerOfMass(model, data_ref, q); |
| 260 |
|
✗ |
Vector3d static_com_ref; |
| 261 |
|
✗ |
static_com_ref << com; |
| 262 |
|
|
|
| 263 |
|
✗ |
Vector3d static_com = data.staticRegressor * phi; |
| 264 |
|
|
|
| 265 |
|
✗ |
BOOST_CHECK(static_com.isApprox(static_com_ref)); |
| 266 |
|
|
} |
| 267 |
|
|
|
| 268 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_body_regressor) |
| 269 |
|
|
{ |
| 270 |
|
|
using namespace Eigen; |
| 271 |
|
|
using namespace pinocchio; |
| 272 |
|
|
|
| 273 |
|
✗ |
Inertia I(Inertia::Random()); |
| 274 |
|
✗ |
Motion v(Motion::Random()); |
| 275 |
|
✗ |
Motion a(Motion::Random()); |
| 276 |
|
|
|
| 277 |
|
✗ |
Force f = I * a + I.vxiv(v); |
| 278 |
|
|
|
| 279 |
|
✗ |
Inertia::Vector6 f_regressor = bodyRegressor(v, a) * I.toDynamicParameters(); |
| 280 |
|
|
|
| 281 |
|
✗ |
BOOST_CHECK(f_regressor.isApprox(f.toVector())); |
| 282 |
|
|
} |
| 283 |
|
|
|
| 284 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_joint_body_regressor) |
| 285 |
|
|
{ |
| 286 |
|
|
using namespace Eigen; |
| 287 |
|
|
using namespace pinocchio; |
| 288 |
|
|
|
| 289 |
|
✗ |
pinocchio::Model model; |
| 290 |
|
✗ |
buildModels::manipulator(model); |
| 291 |
|
✗ |
pinocchio::Data data(model); |
| 292 |
|
|
|
| 293 |
|
✗ |
JointIndex JOINT_ID = JointIndex(model.njoints) - 1; |
| 294 |
|
|
|
| 295 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 296 |
|
✗ |
VectorXd v = Eigen::VectorXd::Random(model.nv); |
| 297 |
|
✗ |
VectorXd a = Eigen::VectorXd::Random(model.nv); |
| 298 |
|
|
|
| 299 |
|
✗ |
rnea(model, data, q, v, a); |
| 300 |
|
|
|
| 301 |
|
✗ |
Force f = data.f[JOINT_ID]; |
| 302 |
|
|
|
| 303 |
|
|
Inertia::Vector6 f_regressor = |
| 304 |
|
✗ |
jointBodyRegressor(model, data, JOINT_ID) * model.inertias[JOINT_ID].toDynamicParameters(); |
| 305 |
|
|
|
| 306 |
|
✗ |
BOOST_CHECK(f_regressor.isApprox(f.toVector())); |
| 307 |
|
|
} |
| 308 |
|
|
|
| 309 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_frame_body_regressor) |
| 310 |
|
|
{ |
| 311 |
|
|
using namespace Eigen; |
| 312 |
|
|
using namespace pinocchio; |
| 313 |
|
|
|
| 314 |
|
✗ |
pinocchio::Model model; |
| 315 |
|
✗ |
buildModels::manipulator(model); |
| 316 |
|
|
|
| 317 |
|
✗ |
JointIndex JOINT_ID = JointIndex(model.njoints) - 1; |
| 318 |
|
|
|
| 319 |
|
✗ |
const SE3 & framePlacement = SE3::Random(); |
| 320 |
|
✗ |
FrameIndex FRAME_ID = model.addBodyFrame("test_body", JOINT_ID, framePlacement, -1); |
| 321 |
|
|
|
| 322 |
|
✗ |
pinocchio::Data data(model); |
| 323 |
|
|
|
| 324 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 325 |
|
✗ |
VectorXd v = Eigen::VectorXd::Random(model.nv); |
| 326 |
|
✗ |
VectorXd a = Eigen::VectorXd::Random(model.nv); |
| 327 |
|
|
|
| 328 |
|
✗ |
rnea(model, data, q, v, a); |
| 329 |
|
|
|
| 330 |
|
✗ |
Force f = framePlacement.actInv(data.f[JOINT_ID]); |
| 331 |
|
✗ |
Inertia I = framePlacement.actInv(model.inertias[JOINT_ID]); |
| 332 |
|
|
|
| 333 |
|
|
Inertia::Vector6 f_regressor = |
| 334 |
|
✗ |
frameBodyRegressor(model, data, FRAME_ID) * I.toDynamicParameters(); |
| 335 |
|
|
|
| 336 |
|
✗ |
BOOST_CHECK(f_regressor.isApprox(f.toVector())); |
| 337 |
|
|
} |
| 338 |
|
|
|
| 339 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_joint_torque_regressor) |
| 340 |
|
|
{ |
| 341 |
|
|
using namespace Eigen; |
| 342 |
|
|
using namespace pinocchio; |
| 343 |
|
|
|
| 344 |
|
✗ |
pinocchio::Model model; |
| 345 |
|
✗ |
buildModels::humanoidRandom(model); |
| 346 |
|
|
|
| 347 |
|
✗ |
model.lowerPositionLimit.head<7>().fill(-1.); |
| 348 |
|
✗ |
model.upperPositionLimit.head<7>().fill(1.); |
| 349 |
|
|
|
| 350 |
|
✗ |
pinocchio::Data data(model); |
| 351 |
|
✗ |
pinocchio::Data data_ref(model); |
| 352 |
|
|
|
| 353 |
|
✗ |
VectorXd q = randomConfiguration(model); |
| 354 |
|
✗ |
VectorXd v = Eigen::VectorXd::Random(model.nv); |
| 355 |
|
✗ |
VectorXd a = Eigen::VectorXd::Random(model.nv); |
| 356 |
|
|
|
| 357 |
|
✗ |
rnea(model, data_ref, q, v, a); |
| 358 |
|
|
|
| 359 |
|
✗ |
Eigen::VectorXd params(10 * (model.njoints - 1)); |
| 360 |
|
✗ |
for (JointIndex i = 1; i < (Model::JointIndex)model.njoints; ++i) |
| 361 |
|
✗ |
params.segment<10>((int)((i - 1) * 10)) = model.inertias[i].toDynamicParameters(); |
| 362 |
|
|
|
| 363 |
|
✗ |
computeJointTorqueRegressor(model, data, q, v, a); |
| 364 |
|
|
|
| 365 |
|
✗ |
Eigen::VectorXd tau_regressor = data.jointTorqueRegressor * params; |
| 366 |
|
|
|
| 367 |
|
✗ |
BOOST_CHECK(tau_regressor.isApprox(data_ref.tau)); |
| 368 |
|
|
} |
| 369 |
|
|
|
| 370 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_kinetic_energy_regressor) |
| 371 |
|
|
{ |
| 372 |
|
|
using namespace Eigen; |
| 373 |
|
|
using namespace pinocchio; |
| 374 |
|
|
|
| 375 |
|
✗ |
pinocchio::Model model; |
| 376 |
|
✗ |
buildModels::humanoidRandom(model); |
| 377 |
|
|
|
| 378 |
|
✗ |
model.lowerPositionLimit.head<7>().fill(-1.); |
| 379 |
|
✗ |
model.upperPositionLimit.head<7>().fill(1.); |
| 380 |
|
|
|
| 381 |
|
✗ |
pinocchio::Data data(model); |
| 382 |
|
✗ |
pinocchio::Data data_ref(model); |
| 383 |
|
|
|
| 384 |
|
✗ |
const VectorXd q = randomConfiguration(model); |
| 385 |
|
✗ |
const VectorXd v = Eigen::VectorXd::Random(model.nv); |
| 386 |
|
|
|
| 387 |
|
✗ |
computeAllTerms(model, data_ref, q, v); |
| 388 |
|
✗ |
auto target_energy = computeKineticEnergy(model, data_ref); |
| 389 |
|
|
|
| 390 |
|
✗ |
const auto regressor = computeKineticEnergyRegressor(model, data, q, v); |
| 391 |
|
|
|
| 392 |
|
✗ |
Eigen::VectorXd params(10 * (model.njoints - 1)); |
| 393 |
|
✗ |
for (JointIndex i = 1; i < (Model::JointIndex)model.njoints; ++i) |
| 394 |
|
✗ |
params.segment<10>(Eigen::DenseIndex((i - 1) * 10)) = model.inertias[i].toDynamicParameters(); |
| 395 |
|
|
|
| 396 |
|
✗ |
const double kinetic_energy_regressor = data.kineticEnergyRegressor * params; |
| 397 |
|
|
|
| 398 |
|
✗ |
BOOST_CHECK_CLOSE(kinetic_energy_regressor, target_energy, 1e-12); |
| 399 |
|
|
} |
| 400 |
|
|
|
| 401 |
|
✗ |
BOOST_AUTO_TEST_CASE(test_potential_energy_regressor) |
| 402 |
|
|
{ |
| 403 |
|
|
using namespace Eigen; |
| 404 |
|
|
using namespace pinocchio; |
| 405 |
|
|
|
| 406 |
|
✗ |
pinocchio::Model model; |
| 407 |
|
✗ |
buildModels::humanoidRandom(model); |
| 408 |
|
|
|
| 409 |
|
✗ |
model.lowerPositionLimit.head<7>().fill(-1.); |
| 410 |
|
✗ |
model.upperPositionLimit.head<7>().fill(1.); |
| 411 |
|
|
|
| 412 |
|
✗ |
pinocchio::Data data(model); |
| 413 |
|
✗ |
pinocchio::Data data_ref(model); |
| 414 |
|
|
|
| 415 |
|
✗ |
const VectorXd q = randomConfiguration(model); |
| 416 |
|
✗ |
const VectorXd v = Eigen::VectorXd::Random(model.nv); |
| 417 |
|
|
|
| 418 |
|
✗ |
computeAllTerms(model, data_ref, q, v); |
| 419 |
|
✗ |
const double target_energy = computePotentialEnergy(model, data_ref); |
| 420 |
|
|
|
| 421 |
|
✗ |
Eigen::VectorXd params(10 * (model.njoints - 1)); |
| 422 |
|
✗ |
for (JointIndex i = 1; i < (Model::JointIndex)model.njoints; ++i) |
| 423 |
|
✗ |
params.segment<10>(Eigen::DenseIndex((i - 1) * 10)) = model.inertias[i].toDynamicParameters(); |
| 424 |
|
|
|
| 425 |
|
✗ |
computePotentialEnergyRegressor(model, data, q); |
| 426 |
|
✗ |
const double potential_energy_regressor = data.potentialEnergyRegressor * params; |
| 427 |
|
|
|
| 428 |
|
✗ |
BOOST_CHECK_CLOSE(potential_energy_regressor, target_energy, 1e-12); |
| 429 |
|
|
} |
| 430 |
|
|
|
| 431 |
|
|
BOOST_AUTO_TEST_SUITE_END() |
| 432 |
|
|
|