| 1 | /* ---------------------------------------------------------------------------- |
| 2 | |
| 3 | * GTSAM Copyright 2010, Georgia Tech Research Corporation, |
| 4 | * Atlanta, Georgia 30332-0415 |
| 5 | * All Rights Reserved |
| 6 | * Authors: Frank Dellaert, et al. (see THANKS for the full author list) |
| 7 | |
| 8 | * See LICENSE for the license information |
| 9 | |
| 10 | * -------------------------------------------------------------------------- */ |
| 11 | |
| 12 | /** |
| 13 | * @file testRFID.cpp |
| 14 | * @brief Unit tests for the RFID factor |
| 15 | * @author Stephen Williams (swilliams8@gatech.edu) |
| 16 | * @date Jan 16, 2012 |
| 17 | */ |
| 18 | |
| 19 | #include <gtsam_unstable/nonlinear/LinearizedFactor.h> |
| 20 | #include <gtsam/base/numericalDerivative.h> |
| 21 | #include <gtsam/geometry/Point3.h> |
| 22 | #include <gtsam/inference/Key.h> |
| 23 | #include <gtsam/nonlinear/NonlinearFactorGraph.h> |
| 24 | #include <gtsam/nonlinear/Values.h> |
| 25 | #include <gtsam/slam/BetweenFactor.h> |
| 26 | #include <CppUnitLite/TestHarness.h> |
| 27 | |
| 28 | using namespace gtsam; |
| 29 | |
| 30 | /* ************************************************************************* */ |
| 31 | TEST( LinearizedFactor, equals_jacobian ) |
| 32 | { |
| 33 | // Create a Between Factor from a Point3. This is actually a linear factor. |
| 34 | Key x1(1); |
| 35 | Key x2(2); |
| 36 | Values values; |
| 37 | values.insert(j: x1, val: Point3(-22.4, +8.5, +2.4)); |
| 38 | values.insert(j: x2, val: Point3(-21.0, +5.0, +21.0)); |
| 39 | |
| 40 | Point3 measured(1.0, -2.5, 17.8); |
| 41 | SharedNoiseModel model = noiseModel::Isotropic::Sigma(dim: 3, sigma: 0.1); |
| 42 | BetweenFactor<Point3> betweenFactor(x1, x2, measured, model); |
| 43 | |
| 44 | |
| 45 | // Create two identical factors and make sure they're equal |
| 46 | JacobianFactor::shared_ptr jf = std::static_pointer_cast<JacobianFactor>(r: betweenFactor.linearize(x: values)); |
| 47 | LinearizedJacobianFactor jacobian1(jf, values); |
| 48 | LinearizedJacobianFactor jacobian2(jf, values); |
| 49 | |
| 50 | CHECK(assert_equal(jacobian1, jacobian2)); |
| 51 | } |
| 52 | |
| 53 | /* ************************************************************************* */ |
| 54 | TEST( LinearizedFactor, clone_jacobian ) |
| 55 | { |
| 56 | // Create a Between Factor from a Point3. This is actually a linear factor. |
| 57 | Key x1(1); |
| 58 | Key x2(2); |
| 59 | Values values; |
| 60 | values.insert(j: x1, val: Point3(-22.4, +8.5, +2.4)); |
| 61 | values.insert(j: x2, val: Point3(-21.0, +5.0, +21.0)); |
| 62 | |
| 63 | Point3 measured(1.0, -2.5, 17.8); |
| 64 | SharedNoiseModel model = noiseModel::Isotropic::Sigma(dim: 3, sigma: 0.1); |
| 65 | BetweenFactor<Point3> betweenFactor(x1, x2, measured, model); |
| 66 | |
| 67 | // Create one factor that is a clone of the other and make sure they're equal |
| 68 | JacobianFactor::shared_ptr jf = std::static_pointer_cast<JacobianFactor>(r: betweenFactor.linearize(x: values)); |
| 69 | LinearizedJacobianFactor jacobian1(jf, values); |
| 70 | LinearizedJacobianFactor::shared_ptr jacobian2 = std::static_pointer_cast<LinearizedJacobianFactor>(r: jacobian1.clone()); |
| 71 | CHECK(assert_equal(jacobian1, *jacobian2)); |
| 72 | |
| 73 | JacobianFactor::shared_ptr jf1 = std::static_pointer_cast<JacobianFactor>(r: jacobian1.linearize(c: values)); |
| 74 | JacobianFactor::shared_ptr jf2 = std::static_pointer_cast<JacobianFactor>(r: jacobian2->linearize(c: values)); |
| 75 | CHECK(assert_equal(*jf1, *jf2)); |
| 76 | |
| 77 | Matrix information1 = jf1->augmentedInformation(); |
| 78 | Matrix information2 = jf2->augmentedInformation(); |
| 79 | CHECK(assert_equal(information1, information2)); |
| 80 | } |
| 81 | |
| 82 | /* ************************************************************************* */ |
| 83 | TEST( LinearizedFactor, add_jacobian ) |
| 84 | { |
| 85 | // Create a Between Factor from a Point3. This is actually a linear factor. |
| 86 | Key x1(1); |
| 87 | Key x2(2); |
| 88 | Values values; |
| 89 | values.insert(j: x1, val: Point3(-22.4, +8.5, +2.4)); |
| 90 | values.insert(j: x2, val: Point3(-21.0, +5.0, +21.0)); |
| 91 | |
| 92 | Point3 measured(1.0, -2.5, 17.8); |
| 93 | SharedNoiseModel model = noiseModel::Isotropic::Sigma(dim: 3, sigma: 0.1); |
| 94 | BetweenFactor<Point3> betweenFactor(x1, x2, measured, model); |
| 95 | |
| 96 | // Create two factor graphs, one using 'push_back' and one using 'add' and make sure they're equal |
| 97 | JacobianFactor::shared_ptr jf = std::static_pointer_cast<JacobianFactor>(r: betweenFactor.linearize(x: values)); |
| 98 | LinearizedJacobianFactor::shared_ptr jacobian(new LinearizedJacobianFactor(jf, values)); |
| 99 | NonlinearFactorGraph graph1; graph1.push_back(factor: jacobian); |
| 100 | NonlinearFactorGraph graph2; graph2.push_back(factor: *jacobian); |
| 101 | |
| 102 | // TODO: When creating a Jacobian from a cached factor, I experienced a problem in the 'add' version |
| 103 | // However, I am currently unable to reproduce the error in this unit test. |
| 104 | // I don't know if this affects the Hessian version as well. |
| 105 | CHECK(assert_equal(graph1, graph2)); |
| 106 | } |
| 107 | |
| 108 | ///* ************************************************************************* */ |
| 109 | //TEST( LinearizedFactor, error_jacobian ) |
| 110 | //{ |
| 111 | // // Create a Between Factor from a Point3. This is actually a linear factor. |
| 112 | // Key key1(1); |
| 113 | // Key key2(2); |
| 114 | // Ordering ordering; |
| 115 | // ordering.push_back(key1); |
| 116 | // ordering.push_back(key2); |
| 117 | // Values values; |
| 118 | // values.insert(key1, Point3(-22.4, +8.5, +2.4)); |
| 119 | // values.insert(key2, Point3(-21.0, +5.0, +21.0)); |
| 120 | // |
| 121 | // Point3 measured(1.0, -2.5, 17.8); |
| 122 | // SharedNoiseModel model = noiseModel::Isotropic::Sigma(3, 0.1); |
| 123 | // BetweenFactor<Point3> betweenFactor(key1, key2, measured, model); |
| 124 | // |
| 125 | // |
| 126 | // // Create a linearized jacobian factors |
| 127 | // JacobianFactor::shared_ptr jf = std::static_pointer_cast<JacobianFactor>(betweenFactor.linearize(values)); |
| 128 | // LinearizedJacobianFactor jacobian(jf, values); |
| 129 | // |
| 130 | // |
| 131 | // for(double x1 = -10; x1 < 10; x1 += 2.0) { |
| 132 | // for(double y1 = -10; y1 < 10; y1 += 2.0) { |
| 133 | // for(double z1 = -10; z1 < 10; z1 += 2.0) { |
| 134 | // |
| 135 | // for(double x2 = -10; x2 < 10; x2 += 2.0) { |
| 136 | // for(double y2 = -10; y2 < 10; y2 += 2.0) { |
| 137 | // for(double z2 = -10; z2 < 10; z2 += 2.0) { |
| 138 | // |
| 139 | // Values linpoint; |
| 140 | // linpoint.insert(key1, Point3(x1, y1, z1)); |
| 141 | // linpoint.insert(key2, Point3(x2, y2, z2)); |
| 142 | // |
| 143 | // // Check that the error of the Linearized Jacobian and the original factor match |
| 144 | // // This only works because a BetweenFactor on a Point3 is actually a linear system |
| 145 | // double expected_error = betweenFactor.error(linpoint); |
| 146 | // double actual_error = jacobian.error(linpoint); |
| 147 | // EXPECT_DOUBLES_EQUAL(expected_error, actual_error, 1e-9 ); |
| 148 | // |
| 149 | // // Check that the linearized factors are identical |
| 150 | // GaussianFactor::shared_ptr expected_factor = betweenFactor.linearize(linpoint); |
| 151 | // GaussianFactor::shared_ptr actual_factor = jacobian.linearize(linpoint); |
| 152 | // CHECK(assert_equal(*expected_factor, *actual_factor)); |
| 153 | // } |
| 154 | // } |
| 155 | // } |
| 156 | // |
| 157 | // } |
| 158 | // } |
| 159 | // } |
| 160 | // |
| 161 | //} |
| 162 | |
| 163 | /* ************************************************************************* */ |
| 164 | TEST( LinearizedFactor, equals_hessian ) |
| 165 | { |
| 166 | // Create a Between Factor from a Point3. This is actually a linear factor. |
| 167 | Key x1(1); |
| 168 | Key x2(2); |
| 169 | Values values; |
| 170 | values.insert(j: x1, val: Point3(-22.4, +8.5, +2.4)); |
| 171 | values.insert(j: x2, val: Point3(-21.0, +5.0, +21.0)); |
| 172 | |
| 173 | Point3 measured(1.0, -2.5, 17.8); |
| 174 | SharedNoiseModel model = noiseModel::Isotropic::Sigma(dim: 3, sigma: 0.1); |
| 175 | BetweenFactor<Point3> betweenFactor(x1, x2, measured, model); |
| 176 | |
| 177 | |
| 178 | // Create two identical factors and make sure they're equal |
| 179 | JacobianFactor::shared_ptr jf = std::static_pointer_cast<JacobianFactor>(r: betweenFactor.linearize(x: values)); |
| 180 | HessianFactor::shared_ptr hf(new HessianFactor(*jf)); |
| 181 | LinearizedHessianFactor hessian1(hf, values); |
| 182 | LinearizedHessianFactor hessian2(hf, values); |
| 183 | |
| 184 | CHECK(assert_equal(hessian1, hessian2)); |
| 185 | } |
| 186 | |
| 187 | /* ************************************************************************* */ |
| 188 | TEST( LinearizedFactor, clone_hessian ) |
| 189 | { |
| 190 | // Create a Between Factor from a Point3. This is actually a linear factor. |
| 191 | Key x1(1); |
| 192 | Key x2(2); |
| 193 | Values values; |
| 194 | values.insert(j: x1, val: Point3(-22.4, +8.5, +2.4)); |
| 195 | values.insert(j: x2, val: Point3(-21.0, +5.0, +21.0)); |
| 196 | |
| 197 | Point3 measured(1.0, -2.5, 17.8); |
| 198 | SharedNoiseModel model = noiseModel::Isotropic::Sigma(dim: 3, sigma: 0.1); |
| 199 | BetweenFactor<Point3> betweenFactor(x1, x2, measured, model); |
| 200 | |
| 201 | |
| 202 | // Create two identical factors and make sure they're equal |
| 203 | JacobianFactor::shared_ptr jf = std::static_pointer_cast<JacobianFactor>(r: betweenFactor.linearize(x: values)); |
| 204 | HessianFactor::shared_ptr hf(new HessianFactor(*jf)); |
| 205 | LinearizedHessianFactor hessian1(hf, values); |
| 206 | LinearizedHessianFactor::shared_ptr hessian2 = std::static_pointer_cast<LinearizedHessianFactor>(r: hessian1.clone()); |
| 207 | |
| 208 | CHECK(assert_equal(hessian1, *hessian2)); |
| 209 | } |
| 210 | |
| 211 | /* ************************************************************************* */ |
| 212 | TEST( LinearizedFactor, add_hessian ) |
| 213 | { |
| 214 | // Create a Between Factor from a Point3. This is actually a linear factor. |
| 215 | Key x1(1); |
| 216 | Key x2(2); |
| 217 | Values values; |
| 218 | values.insert(j: x1, val: Point3(-22.4, +8.5, +2.4)); |
| 219 | values.insert(j: x2, val: Point3(-21.0, +5.0, +21.0)); |
| 220 | |
| 221 | Point3 measured(1.0, -2.5, 17.8); |
| 222 | SharedNoiseModel model = noiseModel::Isotropic::Sigma(dim: 3, sigma: 0.1); |
| 223 | BetweenFactor<Point3> betweenFactor(x1, x2, measured, model); |
| 224 | |
| 225 | |
| 226 | // Create two identical factors and make sure they're equal |
| 227 | JacobianFactor::shared_ptr jf = std::static_pointer_cast<JacobianFactor>(r: betweenFactor.linearize(x: values)); |
| 228 | HessianFactor::shared_ptr hf(new HessianFactor(*jf)); |
| 229 | LinearizedHessianFactor::shared_ptr hessian(new LinearizedHessianFactor(hf, values)); |
| 230 | NonlinearFactorGraph graph1; graph1.push_back(factor: hessian); |
| 231 | NonlinearFactorGraph graph2; graph2.push_back(factor: *hessian); |
| 232 | |
| 233 | CHECK(assert_equal(graph1, graph2)); |
| 234 | } |
| 235 | |
| 236 | ///* ************************************************************************* */ |
| 237 | //TEST( LinearizedFactor, error_hessian ) |
| 238 | //{ |
| 239 | // // Create a Between Factor from a Point3. This is actually a linear factor. |
| 240 | // Key key1(1); |
| 241 | // Key key2(2); |
| 242 | // Ordering ordering; |
| 243 | // ordering.push_back(key1); |
| 244 | // ordering.push_back(key2); |
| 245 | // Values values; |
| 246 | // values.insert(key1, Point3(-22.4, +8.5, +2.4)); |
| 247 | // values.insert(key2, Point3(-21.0, +5.0, +21.0)); |
| 248 | // |
| 249 | // Point3 measured(1.0, -2.5, 17.8); |
| 250 | // SharedNoiseModel model = noiseModel::Isotropic::Sigma(3, 0.1); |
| 251 | // BetweenFactor<Point3> betweenFactor(key1, key2, measured, model); |
| 252 | // |
| 253 | // |
| 254 | // // Create a linearized hessian factor |
| 255 | // JacobianFactor::shared_ptr jf = std::static_pointer_cast<JacobianFactor>(betweenFactor.linearize(values)); |
| 256 | // HessianFactor::shared_ptr hf(new HessianFactor(*jf)); |
| 257 | // LinearizedHessianFactor hessian(hf, values); |
| 258 | // |
| 259 | // |
| 260 | // for(double x1 = -10; x1 < 10; x1 += 2.0) { |
| 261 | // for(double y1 = -10; y1 < 10; y1 += 2.0) { |
| 262 | // for(double z1 = -10; z1 < 10; z1 += 2.0) { |
| 263 | // |
| 264 | // for(double x2 = -10; x2 < 10; x2 += 2.0) { |
| 265 | // for(double y2 = -10; y2 < 10; y2 += 2.0) { |
| 266 | // for(double z2 = -10; z2 < 10; z2 += 2.0) { |
| 267 | // |
| 268 | // Values linpoint; |
| 269 | // linpoint.insert(key1, Point3(x1, y1, z1)); |
| 270 | // linpoint.insert(key2, Point3(x2, y2, z2)); |
| 271 | // |
| 272 | // // Check that the error of the Linearized Hessian and the original factor match |
| 273 | // // This only works because a BetweenFactor on a Point3 is actually a linear system |
| 274 | // double expected_error = betweenFactor.error(linpoint); |
| 275 | // double actual_error = hessian.error(linpoint); |
| 276 | // EXPECT_DOUBLES_EQUAL(expected_error, actual_error, 1e-9 ); |
| 277 | // |
| 278 | // // Check that the linearized factors are identical |
| 279 | // GaussianFactor::shared_ptr expected_factor = HessianFactor::shared_ptr(new HessianFactor(*betweenFactor.linearize(linpoint))); |
| 280 | // GaussianFactor::shared_ptr actual_factor = hessian.linearize(linpoint); |
| 281 | // CHECK(assert_equal(*expected_factor, *actual_factor)); |
| 282 | // } |
| 283 | // } |
| 284 | // } |
| 285 | // |
| 286 | // } |
| 287 | // } |
| 288 | // } |
| 289 | // |
| 290 | //} |
| 291 | |
| 292 | /* ************************************************************************* */ |
| 293 | int main() |
| 294 | { |
| 295 | TestResult tr; return TestRegistry::runAllTests(result&: tr); |
| 296 | } |
| 297 | /* ************************************************************************* */ |
| 298 | |
| 299 | |