The version 1.0 of this SLAM algorithm was designed by Alexander Isenko for the Formula Student competition 2018 for the UAS Munich team - municHMotorsport.

The corresponding master thesis of the LMU Munich can be found here.

If you see this on github, it's only a mirror of our internal municHMotorsport gitlab repository.

We designed this algorithm with performance, future proofness and minimal requirements in mind. Everything is header only besides our own Extended Kalman Filter implementation which has to be installed globally.

It's a c++14 code base with doxygen and inline comments for almost every class, function or member.

Also it's virtual, new, delete and exception free. We use only static allocation for every component to get the O(1) runtime complexity.

We purposefully implemented everything besides the matrix multiplication in the EKF by ourselves to get a deep dive in the algorithms which allows a full performance analysis and possible future improvements without waiting for maintainer outside of the team.

TODO

Install our Kalman Filter library (v1.0). Install our Connector library (v1.0).

> git clone https://github.com/cirquit/clara
> mkdir build && cd build
> cmake .. -DENABLE_OPTIMIZATIONS_CLARA=ON
> make -j4

To enable tests:

> cmake .. -DENABLE_TESTS_CLARA=ON

You can enable critical only debugging (=1), all (=2) or none (=0) with:

> cmake .. -DENABLE_DEBUGGING_LVL_CLARA=2

You can easily reference the headers by hand and link to the clara library in your CMakeLists.txt. If you want a system-level installation just type make install in your build directory.

If you add the following to your own CMakeLists.txt:

find_package(clara version 1.0 REQUIRED)
target_link_libraries(${your-awesome-executable} ${your-awesome-library} clara )

you can reference the library by

#include <clara-1.0/clara.h>

To create documentation install doxygen an run in the source directory:

> doxygen doxygen.config
> cd documentation/latex
> make

Now you can open the PDF at documentation/latex/refman.pdf or the static HTML at documentation/html/index.html in your favourite browser.

We have multiple ipython notebooks and plotting scripts which we used throughout the season which can be found in here. Please read the README for the build tutorial.

• ekf
  • our O(1) extended kalman filter, install the same way as this library (see its README.md)
• connector
  • our own C++ TCP/UDP wrapper, install the same way as this library (see its README.md)
• fast-cpp-csv-parser
  • header only
  • used for tests
  • already included in the source, no need to download anything
• catch
  • header only
  • used for tests as this is a testing framework
  • already included in the source, no need to download anything
• blaze
  • we currently don't have this dependecy, but we already prepared the FindLAPACK.cmake and FindBLAS.cmake to add it if you want to extend it with anything matrix related. Otherwise remove the linking in libraries/CMakeLists.txt

> wget https://bitbucket.org/blaze-lib/blaze/downloads/blaze-3.3.tar.gz
> tar -xvf blaze-3.3.tar.gz
> sudo apt-get install libopenblas-dev
> sudo apt-get install libboost-all-dev
> cmake -DCMAKE_INSTALL_PREFIX=/usr/local/
> sudo make install