Frank
|
3394e85ef7
|
Now allows for a flag to compile Point3 as derived from Vector3
|
2016-02-08 17:34:42 -08:00 |
dellaert
|
27405fc185
|
renaming variables
|
2016-01-28 01:26:57 -08:00 |
Duy-Nguyen Ta
|
bc99c58226
|
fix Rot3-from-Matrix construction for Quaternion case
|
2015-09-17 17:44:00 -04:00 |
Frank Dellaert
|
2deac4b88f
|
Some fixed-size optimizations, some const. Save a quaternion->matrix conversion in transform_from.
|
2015-07-19 18:54:59 -07:00 |
Frank Dellaert
|
6b037ea492
|
bearing test
|
2015-07-11 18:32:10 -07:00 |
Frank Dellaert
|
58a280c672
|
Bearing
|
2015-07-11 16:56:31 -07:00 |
Frank Dellaert
|
ecfa459590
|
Use consistent check and refactored to avoid sqrt
|
2015-07-05 15:40:42 -07:00 |
dellaert
|
22ff9b6aef
|
Made retract etc take Fixed vectors in poses
|
2015-05-26 00:10:03 -07:00 |
Abe
|
13a4da21b2
|
misc bugfixes and cleanup from skydio
|
2015-05-13 22:26:24 -07:00 |
dellaert
|
9955610ea4
|
Back to compile flags
|
2014-12-26 00:19:59 +01:00 |
thduynguyen
|
2aaa7e7564
|
fix degenerated cases in Pose3's ExpmapDerivative. testPose3 passed.
|
2014-12-25 16:59:11 -05:00 |
dellaert
|
47ff09f6c8
|
Pose3 lost its mojo! But there are errors in Expmap/Logmap derivatives around identity...
|
2014-12-25 19:43:32 +01:00 |
dellaert
|
b5c5a268a7
|
Merge remote-tracking branch 'origin/develop' into feature/tighteningTraits
Conflicts:
gtsam/geometry/Pose2.h
gtsam/geometry/Pose3.cpp
gtsam/geometry/tests/testPose3.cpp
|
2014-12-25 18:16:30 +01:00 |
thduynguyen
|
65bbb5e12d
|
use static free function and fix comments
|
2014-12-25 09:07:21 -05:00 |
thduynguyen
|
5ae9f19de2
|
unify duplicated code
|
2014-12-24 14:08:08 -05:00 |
thduynguyen
|
ea80e36b24
|
Pose3 [Expmap/Logmap]Derivative
|
2014-12-24 14:01:16 -05:00 |
dellaert
|
c6ae119414
|
Tightened what is needed for Lie, i.e., fewer versions of retract/localCoordinates
|
2014-12-22 23:42:52 +01:00 |
dellaert
|
99a50a2f87
|
Split up concepts into Group/Manifold/Lie/VectorSpace. Also fixed some implementations and put in exceptions elsewhere, rather than CONCEPT_NOT_IMPLEMENTED macro.
|
2014-12-22 02:50:52 +01:00 |
nsrinivasan7
|
bd6f210b87
|
changed << to =. gives error because range() was removed from PinholeCamera.h ?
|
2014-12-04 09:36:00 -05:00 |
dellaert
|
fcfd232639
|
Added tests, everything checks out
|
2014-12-04 12:28:12 +01:00 |
dellaert
|
cc96529eb6
|
Sanitized Matrix typedefs and constants a la Eigen
|
2014-12-04 10:41:09 +01:00 |
Natesh Srinivasan
|
aad0b2876b
|
Changed Matrix.h to correct return values amd impleemnted rectangular matrix types.
Also changed block operations to <<
|
2014-12-03 15:16:55 -05:00 |
nsrinivasan7
|
deff8b1e25
|
fixed the (*) to -> and code some more code beautification.
|
2014-12-03 11:27:18 -05:00 |
nsrinivasan7
|
7116661a2e
|
changed naming convention of const matrices to _DxD. @dellaert
|
2014-12-03 10:58:09 -05:00 |
Natesh Srinivasan
|
4e557d38e6
|
updated Matrix.h with commonly used matrices.
|
2014-12-03 09:59:10 -05:00 |
nsrinivasan7
|
595afb51fe
|
fixed Rot3(). @dellaert, I will do the '->', Identity(), setZero() etc . once I am fully done with geometry.
|
2014-12-01 18:20:03 -05:00 |
nsrinivasan7
|
1b2d86929a
|
partially fixed Pose3(). Also adressed some of frank's comments from previous commits
|
2014-12-01 15:52:30 -05:00 |
dellaert
|
6505e602d8
|
FixedRef is now OptionalJacobian
|
2014-11-28 17:14:26 +01:00 |
dellaert
|
1e4905ef04
|
Now use new FixedRef type in all methods that I used fixed-sized matrices in to develop Expressions. All copy/paste bloat is now gone, and boost::none arguments are back.
|
2014-11-28 01:56:28 +01:00 |
Richard Roberts
|
aa093a35da
|
Updated all comma initializer usages to use .finished()
|
2014-11-22 16:35:27 -08:00 |
dellaert
|
b21db08ec1
|
Fixed small issue, should not assign to reference in case of Quaternions.
|
2014-11-02 13:03:36 +01:00 |
dellaert
|
c4bf680e96
|
Cached Rot3::transpose(). Inspired by @cbeall3 fix of Unit3, I realized that with one million points and 1000 cameras, the same Matrix3 (R_.transpose()) was computed a 1000 more times than needed. Especially with quaternions this would be expensive, even with Rot3Q.
|
2014-10-15 11:51:41 +02:00 |
dellaert
|
b704b7b1a1
|
Faster versions
|
2014-10-07 19:34:45 +02:00 |
dellaert
|
ea40de6758
|
Fixed Jacobian versions
|
2014-10-06 21:37:05 +02:00 |
dellaert
|
930c77642e
|
Made tarnsform_to derivatives about twice as fast. Biggest culprit is still malloc.
|
2014-10-06 12:12:54 +02:00 |
jing
|
5e9540470a
|
modified Vector_() in gtsam/base and gtsam/geometry folders
|
2014-01-23 01:17:07 -05:00 |
Richard Roberts
|
880d9a8e3c
|
Switched to new Eigen built-in special comma initializer
|
2013-12-16 21:33:12 +00:00 |
Jing Dong
|
e4d56df32e
|
Fix Matrix_(...) to Mat() <<…
|
2013-11-14 05:39:29 +00:00 |
Frank Dellaert
|
752a9877c5
|
Reformatted with new style file, renamed some derivatives to Dcal, Dpose, Dpoint etc.
|
2013-10-12 05:13:36 +00:00 |
Frank Dellaert
|
1251ba9abb
|
ostream << operator
|
2013-06-06 01:21:34 +00:00 |
Duy-Nguyen Ta
|
104ad15e91
|
Separate adjointMap, which returns a matrix, and adjoint+adjointTranspose actions on a vector with optional derivatives. Also, change dExpInv_TLN to dExpInv_exp with higher order of approximation
|
2013-04-29 21:22:33 +00:00 |
Alex Cunningham
|
26f5f93c60
|
Added Pose3::align() for labeled correspondences. Fixed depreciated make targets.
|
2013-04-29 18:06:33 +00:00 |
Duy-Nguyen Ta
|
cde44b2952
|
Finally finish implementing the Reconstruction factor for updating the poses of holonomic vehicles using discrete variational integrators (eq. 10 in [Kobilarov09siggraph]). Also, better approximation for the derivative of the inverse expmap wrt the lie algebra in Pose3. Test with numericalderivative.
|
2013-04-22 08:34:40 +00:00 |
Duy-Nguyen Ta
|
96f9ecaf53
|
Change "adjoint" and "adjoint_map" to Adjoint and Adjoint_map, since they correspond to Lie group's Ad operator. Implement the Lie algebra's adjoint operator, aka Lie bracket, and the inverse right-trivialized tangent map of the exponential map using the trapezoidal Lie-Newmark scheme, as detailed in [Kobilarov09siggraph]
|
2013-04-21 05:50:07 +00:00 |
Frank Dellaert
|
620583f3c3
|
Pose3::between now 4 times faster (in Debug, did not test in Release)
|
2013-04-05 06:44:16 +00:00 |
Chris Beall
|
4297d24c96
|
changed tabs to spaces for consistent indentation in all of GTSAM
|
2012-10-02 14:40:07 +00:00 |
Frank Dellaert
|
747342ffda
|
Changed some printing details, prompted by MATLAB display
|
2012-07-02 15:01:28 +00:00 |
Chris Beall
|
d95e1738d4
|
fixed size improvements for significant speedup in LBA
|
2012-05-09 17:19:01 +00:00 |
Chris Beall
|
1623b8ce12
|
converted Rot3M to fixed-size Matrix, and changed some methods elsewhere to return fixed-size Vector3 to avoid heap allocations for speedup.
|
2012-04-11 06:46:19 +00:00 |
Frank Dellaert
|
c75f7ead65
|
explicit retract versions for calling in MATLAB
|
2012-01-10 20:23:48 +00:00 |
Frank Dellaert
|
659e524fa0
|
Constructor from Pose2
|
2012-01-10 05:05:36 +00:00 |
Richard Roberts
|
f2638b2c0c
|
Fixed names of coordinates modes in Rot3 and Pose3
|
2012-01-08 21:25:29 +00:00 |
Richard Roberts
|
4f59f58cd1
|
retract and localCoordinates optional argument in Pose3 to switch between different math versions, and unit testing all versions
|
2012-01-08 20:43:17 +00:00 |
Frank Dellaert
|
651259e26b
|
Added comment and sped up both Logmap (by a little) and localCoordinates (by more, esp. in combination with unrotate improvement)
|
2012-01-07 21:12:40 +00:00 |
Frank Dellaert
|
e6382c7ec0
|
Fixed major problem with Pose3 retract. All retractions have to have exactly the same derivatives as the exponential map. Hence, it should never make sense to have to match derivaties with a retract version. I deleted all "wrong" derivatives and pushforwards. Finally, CalibratedCamera had to be modified as well as it hardcoded the wrong derivative of transform_to (for efficiency). It now simply implements the chain rule.
|
2012-01-05 16:45:52 +00:00 |
Frank Dellaert
|
53946b28d0
|
Fixed calls of adjointMap and tested correct exponential map derivatives based on them
|
2012-01-04 14:23:50 +00:00 |
Alex Cunningham
|
4b4815e97f
|
Added more functions to wrap, started going though wrap tests
|
2011-12-01 18:57:32 +00:00 |
Alex Cunningham
|
a7ea0f4e04
|
Formatting and comments, adding Rot3 and Pose3 to matlab interface
|
2011-12-01 01:59:34 +00:00 |
Alex Cunningham
|
cc858224aa
|
Comments only - broken parts in Pose3 correct expmap
|
2011-11-26 22:28:45 +00:00 |
Alex Cunningham
|
af9320e468
|
Cleaned up commented code
|
2011-11-10 15:19:04 +00:00 |
Alex Cunningham
|
4284f07a61
|
Updates to dogleg to work with newer Lie/Manifold/Group interface
|
2011-11-05 23:01:50 +00:00 |
Alex Cunningham
|
2b9a3db085
|
Added Group concept, reworked naming and conventions to reduce unnecessary functions
|
2011-11-05 23:01:43 +00:00 |
Alex Cunningham
|
07aaf38245
|
Geometry passes tests with newer interfaces and more concept checks
|
2011-11-04 21:44:29 +00:00 |
Alex Cunningham
|
b22a39e663
|
Cleaned up use of concept checks in generic factors
|
2011-10-15 18:38:51 +00:00 |
Alex Cunningham
|
c6e2547dad
|
Added concept checking for "Poses", and made both Pose2 and Pose3 comply
|
2011-10-06 20:47:30 +00:00 |
Alex Cunningham
|
541c1085db
|
Added a range to pose function for Pose3
|
2011-10-06 20:47:23 +00:00 |
Alex Cunningham
|
e20561be73
|
Merge branch 'eigen'
|
2011-05-20 13:52:08 +00:00 |
Richard Roberts
|
1a94808f57
|
Tightened thresholds in Rot3, Pose3, Rot2, Pose2, that were "correcting" numbers to zero and one with too loose tolerance. This made it necessary to specify large steps for numerical derivatives (1e-2 instead of the default 1e-5). Fixed this, and updated unit tests to use the default step size.
|
2011-04-08 18:41:50 +00:00 |
Alex Cunningham
|
8398c8a53d
|
Made versions with [expmap|logmap]Full() for Pose2 and Pose3 to allow access to complete expmap functions, while also allowing for the concurrent existence of the approximate expmap for optimization speed
|
2011-02-02 05:17:21 +00:00 |
Alex Cunningham
|
120b6878cf
|
Fixed range function for Pose3
|
2011-01-13 17:09:29 +00:00 |
Alex Cunningham
|
6348a33165
|
Added range measurement function to Pose3
|
2011-01-12 19:31:04 +00:00 |
Richard Roberts
|
2c90e3d836
|
Moved the subdirectories that compile into libgtsam.la into a 'gtsam' subdirectory. Hopefully a good fix for the problem we had trying to make a gtsam distribution, that the source directory may not be called 'gtsam'.
|
2010-10-25 20:10:33 +00:00 |