Important Note
As of August 1 2020, the develop branch is officially in "Pre 4.1" mode, and features deprecated in 4.0 have been removed. Please use the last 4.0.3 release if you need those features.
However, most are easily converted and can be tracked down (in 4.0.3) by disabling the cmake flag GTSAM_ALLOW_DEPRECATED_SINCE_V4.
GTSAM is a C++ library that implements smoothing and mapping (SAM) in robotics and vision, using Factor Graphs and Bayes Networks as the underlying computing paradigm rather than sparse matrices.
| Platform | Compiler | Build Status |
|---|---|---|
| Ubuntu 18.04 | gcc/clang |  |
| macOS | clang |  |
| Windows | MSVC |  |
On top of the C++ library, GTSAM includes wrappers for MATLAB & Python.
In the root library folder execute:
#!bash
$ mkdir build
$ cd build
$ cmake ..
$ make check (optional, runs unit tests)
$ make installPrerequisites:
- Boost >= 1.43 (Ubuntu:
sudo apt-get install libboost-all-dev) - CMake >= 3.0 (Ubuntu:
sudo apt-get install cmake) - A modern compiler, i.e., at least gcc 4.7.3 on Linux.
Optional prerequisites - used automatically if findable by CMake:
- Intel Threaded Building Blocks (TBB) (Ubuntu:
sudo apt-get install libtbb-dev) - Intel Math Kernel Library (MKL) (Ubuntu: installing using APT)
- See INSTALL.md for more installation information
- Note that MKL may not provide a speedup in all cases. Make sure to benchmark your problem with and without MKL.
GTSAM 4 introduces several new features, most notably Expressions and a Python toolbox. It also introduces traits, a C++ technique that allows optimizing with non-GTSAM types. That opens the door to retiring geometric types such as Point2 and Point3 to pure Eigen types, which we also do. A significant change which will not trigger a compile error is that zero-initializing of Point2 and Point3 is deprecated, so please be aware that this might render functions using their default constructor incorrect.
GTSAM 4 also deprecated some legacy functionality and wrongly named methods. If you are on a 4.0.X release, you can define the flag GTSAM_ALLOW_DEPRECATED_SINCE_V4 to use the deprecated methods.
GTSAM 4.1 added a new pybind wrapper, and removed the deprecated functionality. There is a flag GTSAM_ALLOW_DEPRECATED_SINCE_V41 for newly deprecated methods since the 4.1 release, which is on by default, allowing anyone to just pull version 4.1 and compile.
We provide support for MATLAB and Python wrappers for GTSAM. Please refer to the linked documents for more details.
GTSAM includes a state of the art IMU handling scheme based on
- Todd Lupton and Salah Sukkarieh, "Visual-Inertial-Aided Navigation for High-Dynamic Motion in Built Environments Without Initial Conditions", TRO, 28(1):61-76, 2012. [link]
Our implementation improves on this using integration on the manifold, as detailed in
- Luca Carlone, Zsolt Kira, Chris Beall, Vadim Indelman, and Frank Dellaert, "Eliminating conditionally independent sets in factor graphs: a unifying perspective based on smart factors", Int. Conf. on Robotics and Automation (ICRA), 2014. [link]
- Christian Forster, Luca Carlone, Frank Dellaert, and Davide Scaramuzza, "IMU Preintegration on Manifold for Efficient Visual-Inertial Maximum-a-Posteriori Estimation", Robotics: Science and Systems (RSS), 2015. [link]
If you are using the factor in academic work, please cite the publications above.
In GTSAM 4 a new and more efficient implementation, based on integrating on the NavState tangent space and detailed in this document, is enabled by default. To switch to the RSS 2015 version, set the flag GTSAM_TANGENT_PREINTEGRATION to OFF.
There is a GTSAM users Google group for general discussion.
Read about important GTSAM-Concepts here. A primer on GTSAM Expressions,
which support (superfast) automatic differentiation,
can be found on the GTSAM wiki on BitBucket.
See the INSTALL file for more detailed installation instructions.
GTSAM is open source under the BSD license, see the LICENSE and LICENSE.BSD files.
Please see the examples/ directory and the USAGE file for examples on how to use GTSAM.
GTSAM was developed in the lab of Frank Dellaert at the Georgia Institute of Technology, with the help of many contributors over the years, see THANKS.