In the project InFuse the Common Data Fusion Framework (CDFF) and the Developer Tools (CDFF Dev) were designed, implemented and tested (Dominguez et al., 2018). At this stage, we are excited to present the features of the final product and to explain its usage to any potentially interested user.
We will also greatly encourage an open discussion on the pros and cons of this framework and the potential future developments. In a nutshell, CDFF contains C++ modules aimed for particular roboticperception tasks (e.g. Model based tracking) on top of well known sensor fusion libraries (e.g. PCL). Naturally, it allows for the development of new modules. Along with these modules, CDFF provides the interfaces that enables an easy interconnection of these, as well as, reusable components for data persistence and execution.
Finally, the tools for development include auto-generated Python bindings for the CDFF C++ modules and software for visualization, testing and benchmarking. InFuse was designed to be independent of any Robotic Control Operating System (RCOS) (e.g. ROS), while taking full advantage of any feature o ered by these (e.g. communication layer, interfaces, inspection tools).
The process of developing a data fusion module and its integration in a RCOS for online use will be explained. As well, the process of evaluating modules oine, using real datasets without the need of an RCOS will be explained.
