Integration of Matlab-based controllers with ROS for autonomous agricultural vehicles

In agriculture automation, the Robot Operating System (ROS) is one of the platforms mainly used for robot software development, thanks to its capability of integration with other platforms. In particular, ROS offers the versatility of implementing control algorithms in different programming languages. The traditional approach is to use directly Python or C++. In this work, the integration between ROS and Matlab has been investigated, to explore the possibility to combine the potentialities of both environments: versatility and re-usability of ROS, and richness of tools, especially in control systems, for Matlab. In particular, this thesis focuses on the comparison of control performance for an autonomous ground vehicle achieved implementing the same control laws, i.e. a PID and a LQR, using different programming languages. In details, C++, Python, Matlab, and C++ obtained from Matlab coder have been investigated, comparing their performance in terms con trajectory tracking and execution time. As simulation environment, the Gazebo simulator, integrated with ROS, has been exploited. The results suggest that Matlab tools and support functions add a great contribution to ROS, simplifying the controller development and also helping to get a better optimized code. Moreover, using C++-compiled Matlab code also allows to significantly reduce the computational burden.