mjhaskell / gnss_utils

Utilities for fusing and simulating raw GNSS measurements, and computing satellite positions from ephemeris. Includes ROS demo node

GNSS Utils

This a library used for handling raw GNSS data, as well as an example ROS node for calculating position, velocity and clock bias with iterated least-squares.

I made this so that I could feel confident that I was properly handling the data before I put this into a much larger project. I would not recommend actually using the output of the ROS node for anything - there are much better ways of doing GNSS positioning, but this should be a pretty good first-principles example.

Building

The main library depends only on Eigen. If you don't want the test suite or the ROS node, you should be able to build the library alone (or add it as a subdirectory of a larger project). Building is the typical cmake - make workflow, or you can just put this in the src directory of a catkin workspace.

• If catkin is detected, then the point_positioning_demo ROS node will be built.
• If gtest is detected, then the test suite will be built.

I have abstracted as much as I can using C++ classes. This results in (what I think is) really clean and understandable code. Projects like RTKLIB are amazing, but the old-style C code is a little difficult for me to read. My intention with releasing this package is that someone else out there will be able to get running faster by modifying my code, as opposed to starting with porting RTKLIB.

THANKS:

I used a lot of code from RTKLIB and gnss-sdr. I mainly used these projects as references and copied a few of their functions into the reference_algorithms.cpp file which I test against.

There is a short LaTeX beamer presentation in the doc folder which contains a derivation of the algorithm I'm using for Iterated least-squares. You should be able to build it with a standard LaTeX toolchain.

TODO:

• GLONASS Support
• UBLOX M8T reading/parsing