bug fix in advanced example. Question: Currently, pose2SLAMOptimizer is in fact useless! Is it needed? Why don't we add more functionalities for it?
parent
6f4c95a65b
commit
9f055fc812
|
@ -55,19 +55,19 @@ initialEstimate.insertPose(x3, gtsamPose2(4.1, 0.1, 0.1));
|
|||
initialEstimate.print('initial estimate');
|
||||
|
||||
%% set up solver, choose ordering and optimize
|
||||
params = NonlinearOptimizationParameters_newDrecreaseThresholds(1e-15, 1e-15);
|
||||
params = gtsamNonlinearOptimizationParameters_newDecreaseThresholds(1e-15, 1e-15);
|
||||
|
||||
ord = graph.orderingCOLAMD(initialEstimate);
|
||||
|
||||
result = pose2SLAMOptimizer(graph,initialEstimate,ord,params);
|
||||
|
||||
result.print('final result');
|
||||
%result = pose2SLAMOptimizer(graph,initialEstimate,ord,params);
|
||||
%result.print('final result');
|
||||
|
||||
% %
|
||||
% % disp('\\\');
|
||||
% %
|
||||
% % %% Optimize using Levenberg-Marquardt optimization with an ordering from colamd
|
||||
% % result = graph.optimize(initialEstimate);
|
||||
% % result.print('final result');
|
||||
result = graph.optimize(initialEstimate);
|
||||
result.print('final result');
|
||||
|
||||
%% Get the corresponding dense matrix
|
||||
ord = graph.orderingCOLAMD(result);
|
||||
|
|
Loading…
Reference in New Issue