| Line |
Branch |
Exec |
Source |
| 1 |
|
|
// Copyright (c) 2014, LAAS-CNRS |
| 2 |
|
|
// Authors: Joseph Mirabel (joseph.mirabel@laas.fr) |
| 3 |
|
|
// |
| 4 |
|
|
|
| 5 |
|
|
// Redistribution and use in source and binary forms, with or without |
| 6 |
|
|
// modification, are permitted provided that the following conditions are |
| 7 |
|
|
// met: |
| 8 |
|
|
// |
| 9 |
|
|
// 1. Redistributions of source code must retain the above copyright |
| 10 |
|
|
// notice, this list of conditions and the following disclaimer. |
| 11 |
|
|
// |
| 12 |
|
|
// 2. Redistributions in binary form must reproduce the above copyright |
| 13 |
|
|
// notice, this list of conditions and the following disclaimer in the |
| 14 |
|
|
// documentation and/or other materials provided with the distribution. |
| 15 |
|
|
// |
| 16 |
|
|
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
| 17 |
|
|
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
| 18 |
|
|
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
| 19 |
|
|
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
| 20 |
|
|
// HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
| 21 |
|
|
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
| 22 |
|
|
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
| 23 |
|
|
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
| 24 |
|
|
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 25 |
|
|
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| 26 |
|
|
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH |
| 27 |
|
|
// DAMAGE. |
| 28 |
|
|
|
| 29 |
|
|
#include "hpp/constraints/static-stability.hh" |
| 30 |
|
|
|
| 31 |
|
|
#include <hpp/pinocchio/device.hh> |
| 32 |
|
|
#include <hpp/pinocchio/joint.hh> |
| 33 |
|
|
#include <hpp/pinocchio/liegroup-element.hh> |
| 34 |
|
|
#include <limits> |
| 35 |
|
|
|
| 36 |
|
|
#include "hpp/constraints/tools.hh" |
| 37 |
|
|
|
| 38 |
|
|
namespace hpp { |
| 39 |
|
|
namespace constraints { |
| 40 |
|
|
|
| 41 |
|
|
using hpp::pinocchio::LiegroupElement; |
| 42 |
|
|
|
| 43 |
|
|
const value_type StaticStability::G = 9.81; |
| 44 |
|
|
const Eigen::Matrix<value_type, 6, 1> StaticStability::Gravity = |
| 45 |
|
|
(Eigen::Matrix<value_type, 6, 1>() << 0, 0, -1, 0, 0, 0).finished(); |
| 46 |
|
|
|
| 47 |
|
✗ |
StaticStability::StaticStability(const std::string& name, |
| 48 |
|
|
const DevicePtr_t& robot, |
| 49 |
|
|
const Contacts_t& contacts, |
| 50 |
|
✗ |
const CenterOfMassComputationPtr_t& com) |
| 51 |
|
|
: DifferentiableFunction(robot->configSize(), robot->numberDof(), |
| 52 |
|
✗ |
contacts.size() + 6, name), |
| 53 |
|
✗ |
robot_(robot), |
| 54 |
|
✗ |
contacts_(contacts), |
| 55 |
|
✗ |
com_(com), |
| 56 |
|
✗ |
phi_(Eigen::Matrix<value_type, 6, Eigen::Dynamic>::Zero(6, |
| 57 |
|
✗ |
contacts.size()), |
| 58 |
|
✗ |
Eigen::Matrix<value_type, 6, Eigen::Dynamic>::Zero( |
| 59 |
|
✗ |
6, contacts.size() * robot->numberDof())), |
| 60 |
|
✗ |
u_(contacts.size()), |
| 61 |
|
✗ |
uMinus_(contacts.size()), |
| 62 |
|
✗ |
v_(contacts.size()), |
| 63 |
|
✗ |
uDot_(contacts.size(), robot->numberDof()), |
| 64 |
|
✗ |
uMinusDot_(contacts.size(), robot->numberDof()), |
| 65 |
|
✗ |
vDot_(contacts.size(), robot->numberDof()), |
| 66 |
|
✗ |
lambdaDot_(robot->numberDof()) { |
| 67 |
|
✗ |
phi_.setSize(2, contacts.size()); |
| 68 |
|
✗ |
Traits<PointCom>::Ptr_t OG = PointCom::create(com); |
| 69 |
|
✗ |
for (std::size_t i = 0; i < contacts.size(); ++i) { |
| 70 |
|
|
Traits<PointInJoint>::Ptr_t OP2 = PointInJoint::create( |
| 71 |
|
✗ |
contacts[i].joint, contacts[i].point, robot->numberDof()); |
| 72 |
|
|
Traits<VectorInJoint>::Ptr_t n2 = VectorInJoint::create( |
| 73 |
|
✗ |
contacts[i].joint, contacts[i].normal, robot->numberDof()); |
| 74 |
|
|
|
| 75 |
|
✗ |
phi_(0, i) = n2; |
| 76 |
|
✗ |
phi_(1, i) = (OG - OP2) ^ n2; |
| 77 |
|
|
} |
| 78 |
|
|
} |
| 79 |
|
|
|
| 80 |
|
✗ |
StaticStabilityPtr_t StaticStability::create( |
| 81 |
|
|
const std::string& name, const DevicePtr_t& robot, |
| 82 |
|
|
const Contacts_t& contacts, const CenterOfMassComputationPtr_t& com) { |
| 83 |
|
✗ |
return StaticStabilityPtr_t(new StaticStability(name, robot, contacts, com)); |
| 84 |
|
|
} |
| 85 |
|
|
|
| 86 |
|
✗ |
StaticStabilityPtr_t StaticStability::create( |
| 87 |
|
|
const DevicePtr_t& robot, const Contacts_t& contacts, |
| 88 |
|
|
const CenterOfMassComputationPtr_t& com) { |
| 89 |
|
✗ |
return create("StaticStability", robot, contacts, com); |
| 90 |
|
|
} |
| 91 |
|
|
|
| 92 |
|
✗ |
void StaticStability::impl_compute(LiegroupElementRef result, |
| 93 |
|
|
ConfigurationIn_t argument) const { |
| 94 |
|
✗ |
robot_->currentConfiguration(argument); |
| 95 |
|
✗ |
robot_->computeForwardKinematics(pinocchio::JOINT_POSITION); |
| 96 |
|
|
|
| 97 |
|
✗ |
phi_.invalidate(); |
| 98 |
|
|
|
| 99 |
|
✗ |
phi_.computeSVD(argument); |
| 100 |
|
|
|
| 101 |
|
✗ |
const Eigen::Matrix<value_type, 6, 1> G = -1 * Gravity; |
| 102 |
|
✗ |
u_.noalias() = phi_.svd().solve(G); |
| 103 |
|
|
|
| 104 |
|
✗ |
if (computeUminusAndV(u_, uMinus_, v_)) { |
| 105 |
|
|
// value_type lambda, unused_lMax; size_type iMax, iMin; |
| 106 |
|
|
// findBoundIndex (u_, v_, lambda, &iMin, unused_lMax, &iMax); |
| 107 |
|
✗ |
value_type lambda = 1; |
| 108 |
|
|
|
| 109 |
|
✗ |
result.vector().segment(0, contacts_.size()) = u_ + lambda * v_; |
| 110 |
|
|
} else { |
| 111 |
|
✗ |
result.vector().segment(0, contacts_.size()) = u_; |
| 112 |
|
|
} |
| 113 |
|
✗ |
result.vector().segment<6>(contacts_.size()) = Gravity + phi_.value() * u_; |
| 114 |
|
|
} |
| 115 |
|
|
|
| 116 |
|
✗ |
void StaticStability::impl_jacobian(matrixOut_t jacobian, |
| 117 |
|
|
ConfigurationIn_t argument) const { |
| 118 |
|
✗ |
robot_->currentConfiguration(argument); |
| 119 |
|
✗ |
robot_->computeForwardKinematics(pinocchio::JOINT_POSITION | |
| 120 |
|
|
pinocchio::JACOBIAN); |
| 121 |
|
|
|
| 122 |
|
✗ |
phi_.invalidate(); |
| 123 |
|
|
|
| 124 |
|
✗ |
phi_.computeSVD(argument); |
| 125 |
|
✗ |
phi_.computeJacobian(argument); |
| 126 |
|
✗ |
phi_.computePseudoInverse(argument); |
| 127 |
|
|
|
| 128 |
|
✗ |
const Eigen::Matrix<value_type, 6, 1> G = -1 * Gravity; |
| 129 |
|
✗ |
u_.noalias() = phi_.svd().solve(G); |
| 130 |
|
✗ |
phi_.computePseudoInverseJacobian(argument, G); |
| 131 |
|
✗ |
uDot_.noalias() = phi_.pinvJacobian(); |
| 132 |
|
|
|
| 133 |
|
✗ |
jacobian.block(0, 0, contacts_.size(), robot_->numberDof()).noalias() = uDot_; |
| 134 |
|
|
|
| 135 |
|
✗ |
if (computeUminusAndV(u_, uMinus_, v_)) { |
| 136 |
|
✗ |
matrix_t S = -matrix_t::Identity(u_.size(), u_.size()); |
| 137 |
|
✗ |
S.diagonal() = 1 * (u_.array() >= 0).select(0, -vector_t::Ones(u_.size())); |
| 138 |
|
|
|
| 139 |
|
|
// value_type lambda, unused_lMax; size_type iMax, iMin; |
| 140 |
|
|
// findBoundIndex (u_, v_, lambda, &iMin, unused_lMax, &iMax); |
| 141 |
|
|
// if (lambda < Eigen::NumTraits <value_type>::dummy_precision()) |
| 142 |
|
|
// return; |
| 143 |
|
✗ |
value_type lambda = 1; |
| 144 |
|
|
|
| 145 |
|
✗ |
computeVDot(argument, uMinus_, S.diagonal(), uDot_, uMinusDot_, vDot_); |
| 146 |
|
|
|
| 147 |
|
|
// computeLambdaDot (u_, v_, iMin, uDot_, vDot_, lambdaDot_); |
| 148 |
|
|
|
| 149 |
|
|
// jacobian.block (0, 0, contacts_.size(), robot_->numberDof()).noalias () |
| 150 |
|
|
// += lambda * vDot_ + v_ * lambdaDot_.transpose (); |
| 151 |
|
|
|
| 152 |
|
✗ |
jacobian.block(0, 0, contacts_.size(), robot_->numberDof()).noalias() += |
| 153 |
|
✗ |
lambda * vDot_; |
| 154 |
|
|
} |
| 155 |
|
|
|
| 156 |
|
✗ |
phi_.jacobianTimes( |
| 157 |
|
✗ |
argument, u_, |
| 158 |
|
✗ |
jacobian.block(contacts_.size(), 0, 6, robot_->numberDof())); |
| 159 |
|
✗ |
phi_.computePseudoInverseJacobian(argument, Gravity); |
| 160 |
|
✗ |
jacobian.block(contacts_.size(), 0, 6, robot_->numberDof()) += |
| 161 |
|
✗ |
-phi_.value() * phi_.pinvJacobian(); |
| 162 |
|
|
} |
| 163 |
|
|
|
| 164 |
|
✗ |
void StaticStability::findBoundIndex(vectorIn_t u, vectorIn_t v, |
| 165 |
|
|
value_type& lambdaMin, size_type* iMin, |
| 166 |
|
|
value_type& lambdaMax, size_type* iMax) { |
| 167 |
|
|
// Be carefull when v has small values. |
| 168 |
|
|
// Consider them as 0 |
| 169 |
|
✗ |
const value_type eps = Eigen::NumTraits<value_type>::dummy_precision(); |
| 170 |
|
|
vector_t lambdas = |
| 171 |
|
✗ |
(v.array().cwiseAbs() < eps).select(0, -u.cwiseQuotient(v)); |
| 172 |
|
✗ |
lambdaMin = (v.array() > eps).select(lambdas, 0).maxCoeff(iMin); |
| 173 |
|
✗ |
lambdaMax = (v.array() < eps).select(lambdas, 0).minCoeff(iMax); |
| 174 |
|
|
} |
| 175 |
|
|
|
| 176 |
|
✗ |
bool StaticStability::computeUminusAndV(vectorIn_t u, vectorOut_t uMinus, |
| 177 |
|
|
vectorOut_t v) const { |
| 178 |
|
|
using namespace hpp::pinocchio; |
| 179 |
|
|
|
| 180 |
|
✗ |
uMinus.noalias() = (u.array() >= 0).select(0, -u); |
| 181 |
|
|
|
| 182 |
|
✗ |
if (uMinus.isZero()) return false; |
| 183 |
|
|
|
| 184 |
|
✗ |
size_type rank = phi_.svd().rank(); |
| 185 |
|
✗ |
v.noalias() = getV2<MoE_t::SVD_t>(phi_.svd(), rank) * |
| 186 |
|
✗ |
(getV2<MoE_t::SVD_t>(phi_.svd(), rank).adjoint() * uMinus); |
| 187 |
|
|
// v.noalias() = uMinus; |
| 188 |
|
|
// v.noalias() -= getV1 <MoE_t::SVD_t> (phi_.svd()) * |
| 189 |
|
|
// ( getV1 <MoE_t::SVD_t> (phi_.svd()).adjoint() * uMinus ); |
| 190 |
|
✗ |
return true; |
| 191 |
|
|
} |
| 192 |
|
|
|
| 193 |
|
✗ |
void StaticStability::computeVDot(const ConfigurationIn_t arg, |
| 194 |
|
|
vectorIn_t uMinus, vectorIn_t S, |
| 195 |
|
|
matrixIn_t uDot, matrixOut_t uMinusDot, |
| 196 |
|
|
matrixOut_t vDot) const { |
| 197 |
|
|
using namespace hpp::pinocchio; |
| 198 |
|
|
|
| 199 |
|
✗ |
size_type rank = phi_.svd().rank(); |
| 200 |
|
✗ |
uMinusDot.noalias() = S.asDiagonal() * uDot; |
| 201 |
|
✗ |
vDot.noalias() = uMinusDot; |
| 202 |
|
✗ |
vDot.noalias() -= |
| 203 |
|
✗ |
getV1<MoE_t::SVD_t>(phi_.svd(), rank) * |
| 204 |
|
✗ |
(getV1<MoE_t::SVD_t>(phi_.svd(), rank).adjoint() * uMinusDot); |
| 205 |
|
|
|
| 206 |
|
|
// TODO: preallocate this matrix |
| 207 |
|
|
Eigen::Matrix<value_type, 6, Eigen::Dynamic> JphiTimesUMinus( |
| 208 |
|
✗ |
6, robot_->numberDof()); |
| 209 |
|
✗ |
phi_.jacobianTimes(arg, uMinus, JphiTimesUMinus); |
| 210 |
|
✗ |
vDot.noalias() -= phi_.pinv() * JphiTimesUMinus; |
| 211 |
|
|
|
| 212 |
|
✗ |
phi_.computePseudoInverseJacobian(arg, phi_.value() * uMinus); |
| 213 |
|
✗ |
vDot.noalias() -= phi_.pinvJacobian(); |
| 214 |
|
|
} |
| 215 |
|
|
|
| 216 |
|
✗ |
void StaticStability::computeLambdaDot(vectorIn_t u, vectorIn_t v, |
| 217 |
|
|
const std::size_t i0, matrixIn_t uDot, |
| 218 |
|
|
matrixIn_t vDot, |
| 219 |
|
|
vectorOut_t lambdaDot) const { |
| 220 |
|
✗ |
if (std::abs(v(i0)) < Eigen::NumTraits<value_type>::dummy_precision()) |
| 221 |
|
✗ |
lambdaDot.setZero(); |
| 222 |
|
|
else { |
| 223 |
|
✗ |
lambdaDot.noalias() = -uDot.row(i0) / v(i0); |
| 224 |
|
✗ |
lambdaDot.noalias() += (u(i0) / (v(i0) * v(i0))) * vDot.row(i0); |
| 225 |
|
|
} |
| 226 |
|
|
} |
| 227 |
|
|
} // namespace constraints |
| 228 |
|
|
} // namespace hpp |
| 229 |
|
|
|