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

added debug info for gradient code

release/4.3a0
Luca 2014-09-06 18:27:07 -04:00
parent 41bb99b48a
commit e3ec28ebeb
1 changed files with 12 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
gtsam/slam/InitializePose3.cpp
View File

@ -185,7 +185,7 @@ Values computeOrientationsGradient(const NonlinearFactorGraph& pose3Graph, const
double stepsize = 2/mu_max; // = 1/(a b dG) double stepsize = 2/mu_max; // = 1/(a b dG)
std::cout <<" b " << b <<" f0 " << f0 <<" a " << a <<" rho " << rho <<" stepsize " << stepsize << " maxNodeDeg "<< maxNodeDeg << std::endl; std::cout <<" b " << b <<" f0 " << f0 <<" a " << a <<" rho " << rho <<" stepsize " << stepsize << " maxNodeDeg "<< maxNodeDeg << std::endl;
double maxGrad;
// gradient iterations // gradient iterations
size_t it; size_t it;
for(it=0; it < maxIter; it++){ for(it=0; it < maxIter; it++){
@ -193,7 +193,7 @@ Values computeOrientationsGradient(const NonlinearFactorGraph& pose3Graph, const
// compute the gradient at each node // compute the gradient at each node
//std::cout << "it " << it <<" b " << b <<" f0 " << f0 <<" a " << a //std::cout << "it " << it <<" b " << b <<" f0 " << f0 <<" a " << a
// <<" rho " << rho <<" stepsize " << stepsize << " maxNodeDeg "<< maxNodeDeg << std::endl; // <<" rho " << rho <<" stepsize " << stepsize << " maxNodeDeg "<< maxNodeDeg << std::endl;
double maxGrad = 0; maxGrad = 0;
BOOST_FOREACH(const Values::ConstKeyValuePair& key_value, inverseRot) { BOOST_FOREACH(const Values::ConstKeyValuePair& key_value, inverseRot) {
Key key = key_value.key; Key key = key_value.key;
//std::cout << "---------------------------key " << DefaultKeyFormatter(key) << std::endl; //std::cout << "---------------------------key " << DefaultKeyFormatter(key) << std::endl;
@ -234,7 +234,7 @@ Values computeOrientationsGradient(const NonlinearFactorGraph& pose3Graph, const
break; break;
} // enf of gradient iterations } // enf of gradient iterations
std::cout << "nr of gradient iterations " << it << std::endl; std::cout << "nr of gradient iterations " << it << "maxGrad " << maxGrad << std::endl;
// Return correct rotations // Return correct rotations
const Rot3& Rref = inverseRot.at<Rot3>(keyAnchor); // This will be set to the identity as so far we included no prior const Rot3& Rref = inverseRot.at<Rot3>(keyAnchor); // This will be set to the identity as so far we included no prior
@ -287,13 +287,13 @@ Vector3 gradientTron(const Rot3& R1, const Rot3& R2, const double a, const doubl
Vector3 logRot = Rot3::Logmap(R1.between(R2)); Vector3 logRot = Rot3::Logmap(R1.between(R2));
double th = logRot.norm(); double th = logRot.norm();
if(th < 1e-4 || th != th){ // the second case means that th = nan (logRot does not work well for +/-pi) if(th != th){ // the second case means that th = nan (logRot does not work well for +/-pi)
Rot3 R1pert = R1.compose( Rot3::Expmap((Vector(3)<< 0.01, 0.01, 0.01)) ); // some perturbation Rot3 R1pert = R1.compose( Rot3::Expmap((Vector(3)<< 0.01, 0.01, 0.01)) ); // some perturbation
logRot = Rot3::Logmap(R1pert.between(R2)); logRot = Rot3::Logmap(R1pert.between(R2));
th = logRot.norm(); th = logRot.norm();
} }
// exclude small or invalid rotations // exclude small or invalid rotations
if (th > 1e-5 || th == th){ // the second case means that th = nan (logRot does not work well for +/-pi) if (th > 1e-5 && th == th){ // nonzero valid rotations
logRot = logRot / th; logRot = logRot / th;
}else{ }else{
logRot = Vector3::Zero(); logRot = Vector3::Zero();
@ -334,7 +334,13 @@ Values computePoses(NonlinearFactorGraph& pose3graph, Values& initialRot) {
// Create optimizer // Create optimizer
GaussNewtonParams params; GaussNewtonParams params;
params.maxIterations = 1; bool singleIter = false;
if(singleIter){
params.maxIterations = 1;
}else{
std::cout << " \n\n\n\n performing more than 1 GN iterations \n\n\n" <<std::endl;
params.setVerbosity("TERMINATION");
}
GaussNewtonOptimizer optimizer(pose3graph, initialPose, params); GaussNewtonOptimizer optimizer(pose3graph, initialPose, params);
Values GNresult = optimizer.optimize(); Values GNresult = optimizer.optimize();