12345qiupeng
/
gtsam
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Pose2Constraint, initial check in with working linearize

release/4.3a0
Frank Dellaert 2009-12-08 20:48:39 +00:00
parent 16da0895a8
commit 066b80e02e
2 changed files with 100 additions and 45 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
cpp/testMatrix.cpp
View File

@ -591,6 +591,24 @@ TEST( matrix, weighted_elimination )
} }
} }
/* ************************************************************************* */
TEST( matrix, inverse_square_root )
{
Matrix measurement_covariance = Matrix_(3,3,
0.25, 0.0, 0.0,
0.0, 0.25, 0.0,
0.0, 0.0, 0.01
);
Matrix actual = inverse_square_root(measurement_covariance);
Matrix square_root_inverse_covariance = Matrix_(3,3,
-2.0, 0.0, 0.0,
0.0, -2.0, 0.0,
0.0, 0.0, -10.0
);
EQUALITY(square_root_inverse_covariance,actual);
}
/* ************************************************************************* */ /* ************************************************************************* */
int main() { TestResult tr; return TestRegistry::runAllTests(tr); } int main() { TestResult tr; return TestRegistry::runAllTests(tr); }

127
cpp/testNonlinearFactor.cpp
View File

@ -14,7 +14,7 @@
#include "Matrix.h" #include "Matrix.h"
#include "smallExample.h" #include "smallExample.h"
#include "Simulated2DMeasurement.h" #include "Simulated2DMeasurement.h"
#include "Pose2.h" #include "GaussianFactor.h"
using namespace std; using namespace std;
using namespace gtsam; using namespace gtsam;
@ -174,59 +174,96 @@ TEST( NonlinearFactor, size )
/* ************************************************************************* */ /* ************************************************************************* */
#include "numericalDerivative.h" #include <map>
#include "Pose2.h"
class Pose2Config : public map<string,Pose2> {
public:
Pose2 get(string key) const {
map<string,Pose2>::const_iterator it = find(key);
if (it==end()) throw std::invalid_argument("invalid key");
return it->second;
}
};
Pose2 h(const Pose2& p1, const Pose2& p2) { class Pose2Constraint : public Factor<Pose2Config> {
Pose2 dpose = p2 - p1; private:
return dpose.rotate(-p1.theta()); std::string key1_, key2_; /** The keys of the two poses, order matters */
} Pose2 measured_;
Matrix square_root_inverse_covariance_; /** sqrt(inv(measurement_covariance)) */
Matrix H1(const Pose2& p1, const Pose2& p2) { public:
double dx= p2.x()-p1.x(); Pose2Constraint(const string& key1, const string& key2,
double dy= p2.y()-p1.y(); const Pose2& measured, const Matrix& measurement_covariance): key1_(key1),key2_(key2),measured_(measured) {
double ct=cos(p1.theta()); square_root_inverse_covariance_ = inverse_square_root(measurement_covariance);
double st=sin(p1.theta()); }
return Matrix_(3,3,
-ct, -st, -st*dx+ct*dy,
st, -ct, -ct*dx-st*dy,
0.0, 0.0, -1.0
);
}
Matrix H2(const Pose2& p1) { /** implement functions needed for Testable */
double ct=cos(p1.theta()); void print(const std::string& name) const {}
double st=sin(p1.theta()); bool equals(const Factor<Pose2Config>& expected, double tol) const {return false;}
return Matrix_(3,3,
ct, st, 0.0,
-st, ct, 0.0,
0.0, 0.0, 1.0
);
}
TEST( PoseConstraintFactor2, testFunctions ) /** implement functions needed to derive from Factor */
double error(const Pose2Config& c) const {return 0;}
std::list<std::string> keys() const { list<string> l; return l; }
std::size_t size() const { return 2;}
/** linearize */
boost::shared_ptr<GaussianFactor> linearize(const Pose2Config& config) const {
Pose2 p1 = config.get(key1_), p2 = config.get(key2_);
Vector b = (measured_ - between(p1,p2)).vector();
Matrix H1 = Dbetween1(p1,p2);
Matrix H2 = Dbetween2(p1,p2);
boost::shared_ptr<GaussianFactor> linearized(new GaussianFactor(
key1_, square_root_inverse_covariance_ * H1,
key2_, square_root_inverse_covariance_ * H2,
b, 1.0));
return linearized;
}
};
TEST( Pose2Constraint, constructor )
{ {
Pose2 p1(0.0, 6.0, 0.0); // create a factor between unknown poses p1 and p2
Pose2 p2(0.101826, 6.111236, 0.011499); Pose2 measured(2,2,M_PI_2);
Matrix measurement_covariance = Matrix_(3,3,
0.25, 0.0, 0.0,
0.0, 0.25, 0.0,
0.0, 0.0, 0.01
);
Pose2Constraint constraint("p1","p2",measured, measurement_covariance);
//expected // Choose a linearization point
Pose2 expectedh(0.101826, 0.111236, 0.011499); Pose2 p1(1.1,2,M_PI_2); // robot at (1.1,2) looking towards y (ground truth is at 1,2, see testPose2)
Matrix expectedH1 = Matrix_(3,3,-1.0, 0.0, 0.111236, 0.0, -1.0, -0.101826, 0.0, 0.0, -1.0); Pose2 p2(-1,4.1,M_PI); // robot at (-1,4) looking at negative (ground truth is at 4.1,2)
Matrix expectedH2 = Matrix_(3,3, 1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0); Pose2Config config;
config.insert(make_pair("p1",p1));
config.insert(make_pair("p2",p2));
// actual // Linearize
Pose2 actualh = h(p1,p2); boost::shared_ptr<GaussianFactor> actual = constraint.linearize(config);
Matrix actualH1 = H1(p1,p2);
Matrix actualH2 = H2(p1);
CHECK(assert_equal(expectedh,actualh)); // expected
CHECK(assert_equal(expectedH1,actualH1)); Matrix expectedH1 = Matrix_(3,3,
CHECK(assert_equal(expectedH2,actualH2)); 0.0,-1.0,2.1,
1.0,0.0,-2.1,
0.0,0.0,-1.0
);
Matrix expectedH2 = Matrix_(3,3,
0.0,1.0,0.0,
-1.0,0.0,0.0,
0.0,0.0,1.0
);
// we need the minus signs as inverse_square_root uses SVD and sign is simply arbitrary (still a ssquare root!)
Matrix square_root_inverse_covariance = Matrix_(3,3,
-2.0, 0.0, 0.0,
0.0, -2.0, 0.0,
0.0, 0.0, -10.0
);
GaussianFactor expected(
"p1", square_root_inverse_covariance*expectedH1,
"p2", square_root_inverse_covariance*expectedH2,
Vector_(3,-0.1,-0.1,0.0), 1.0);
Matrix numericalH1 = numericalDerivative21(h, p1, p2, 1e-5); CHECK(assert_equal(expected,*actual));
//CHECK(assert_equal(numericalH1,actualH1));
Matrix numericalH2 = numericalDerivative22(h, p1, p2, 1e-5);
CHECK(assert_equal(numericalH2,actualH2));
} }
/* ************************************************************************* */ /* ************************************************************************* */