ac10c440e1
PairConfig is implemented, VSLAMConfig is now a typedef!
2010-01-14 02:58:29 +00:00
6b3e8cf49c
simulated2D now reduced to one .h and .cpp, in its own namespace, better naming, and new-style functions to serve as example
2010-01-14 02:50:06 +00:00
93465945e9
Large gtsam refactoring
...
To support faster development *and* better performance Richard and I pushed through a large refactoring of NonlinearFactors.
The following are the biggest changes:
1) NonLinearFactor1 and NonLinearFactor2 are now templated on Config, Key type, and X type, where X is the argument to the measurement function.
2) The measurement itself is no longer kept in the nonlinear factor. Instead, a derived class (see testVSLAMFactor, testNonlinearEquality, testPose3Factor etc...) has to implement a function to compute the errors, "evaluateErrors". Instead of (h(x)-z), it needs to return (z-h(x)), so Ax-b is an approximation of the error. IMPORTANT: evaluateErrors needs - if asked - *combine* the calculation of the function value h(x) and the derivatives dh(x)/dx. This was a major performance issue. To do this, boost::optional<Matrix&> arguments are provided, and tin EvaluateErrors you just says something like
if (H) *H = Matrix_(3,6,....);
3) We are no longer using int or strings for nonlinear factors. Instead, the preferred key type is now Symbol, defined in Key.h. This is both fast and cool: you can construct it from an int, and cast it to a strong. It also does type checking: a Symbol<Pose3,'x'> will not match a Symbol<Pose2,'x'>
4) minor: take a look at LieConfig.h: it help you avoid writing a lot of code bu automatically creating configs for a certain type. See e.g. Pose3Config.h. A "double" LieConfig is on the way - Thanks Richard and Manohar !
2010-01-13 22:25:03 +00:00
2c8d8dbde4
NonlinearOptimizer constructor now requires shared_ptr to prevent storing pointers to stack and temporary variables. Code that uses it will need to be modified, but CitySLAM is already done.
2010-01-11 20:17:28 +00:00
19a3e228d7
Large refactoring - made several Lie group functions global, which used to be member functions, to treat Lie groups more uniformly. Also created Lie.h, and a preprocessor flag in numericalDerivative to change the coordinate frame derivatives are reported in. gtsam and easylib build and pass unit tests, but this will probably break other projects, which will require a few small changes to work again. Email coming in a few minutes to describe the changes.
2010-01-08 00:40:17 +00:00
e2bc13a2a6
Renaming gradients -> jacobians
2009-12-18 03:05:47 +00:00
a1918056a5
Changed NonlinearConstraints to take cost and gradient functions that do not take the list of keys. Tests have been reconstructed using boost::bind
2009-12-18 02:39:02 +00:00
7d0de77fc6
Changed argument ordering for constraints
2009-12-18 01:40:29 +00:00
9efac7b3fb
Some re-naming and re-formatting only
2009-12-15 05:34:49 +00:00
a3ce3f31c8
Cleanup with NonlinearConstraints to make the active() function do thresholding for inequality constraints, rather than constraint function itself.
...
testSQP now has all tests active and passing.
Added greaterThanOrEqual() for vector comparison.
2009-11-30 17:36:34 +00:00
395e4ae3f1
Fixed sign issue on nonlinear constraints, so that that it is not necessary to flip the sign on the delta configs before using exmap.
2009-11-28 21:00:09 +00:00
0ff7e3a5d9
Changed interface on constraint g(x) and grad_g(x) functions to take a list of keys, so that all of the variables in a factor can be used as necessary. Moved g(x) into base NonlinearConstraint class and some cleanup.
2009-11-28 19:18:02 +00:00
43e4c66407
Added more detail to VSLAM test that diverges.
2009-11-24 16:03:57 +00:00
e3e79e3888
Added lagrange access function to the SQPOptimizer
...
Added a test testSQP to try changing the initial Lagrange multiplier values to find a way around the stability problem, but there is no effect.
2009-11-24 14:44:46 +00:00
140e79572f
Added a test to testSQP to check VSLAM solution stability when the initial config is away from the ground truth. The test without the nonlinear constraints has the same stability problem, and both the constrained and unconstrained versions are in place and disabled.
2009-11-24 14:14:03 +00:00
fec280db8e
Added a nonlinear constraint version of the stereo SQP example, but the test when the initial configuration is incorrect is not stable and the points continue moving away indefinitely.
2009-11-24 06:39:27 +00:00
36877532ad
Added ground truth for VSLAM example in testSQP
2009-11-23 16:45:53 +00:00
1fd0404ec9
Cleanup in testSQP, also demoed the unary constraint, as well.
2009-11-21 19:05:24 +00:00
a0a250750b
compare config
2009-11-21 04:22:59 +00:00
08f5b074a0
Added verbosity selector for the SQP demo to remove printouts every time it runs
2009-11-20 14:29:11 +00:00
5f848f272b
SQP demo that moves maps into the correct reference frames is now working using the NonlinearConstraint machinery.
2009-11-20 14:05:21 +00:00
6aba2f1c1c
Changed initial SQP example to use exmap using the new scaling function for VectorConfig
2009-11-20 05:13:32 +00:00
ee4a066275
Implemented linearization and equality for unary NonlinearConstraints. Current tests use a scalar example. Split out implementation into a separate implementation file.
2009-11-20 03:04:49 +00:00
429f27550c
Added a NonlinearConstraint and testNonlinearConstraint. There is currently an abstract base class for constraints and a partially implemented unary constraint.
2009-11-19 16:50:18 +00:00
f98f49381a
Added an example to testSQP to use for combining graphs
2009-11-18 21:57:59 +00:00
d700cd2cac
Made some small changes to testSQP to simplify the system and improve convergence.
2009-11-18 20:23:07 +00:00
7fc68e2d90
Updated testSQP to use a real SQP implementation as separate factors on the previous problem. SQP now works.
2009-11-18 19:42:14 +00:00
20830a1534
Added testSQP demo to show a simple problem solved by Sequential Quadratic Programming. To see the outputs, set verbose=true at the start of the test.
2009-11-16 19:12:56 +00:00
Richard Roberts
Frank Dellaert
Alex Cunningham