Communication methods in modern robotic systems are connecting multiple sensors, actuators, micro-controllers and embedded PCs. A lot of development time is used to set up communication between and within those systems. Modern robot control middlewares like e.g. ROS, Orocos and others improving this situation by e.g. using a central name server, which provides addresses of the desired data or the data itself. But they are also introducing single points of failure: If this server is unreachable, no new communication can be set up.
This work (Funded by the Space Agency of the German Aerospace Center (DLR) with federal funds of the Federal Ministry of Economics and Technology (BMWi) in accordance with the parliamentary resolution of the German Parliament, grant no. 50RA1024 and 50RA1025) is focused to provide a fault tolerant communication middleware that is usable in networks with heterogeneous hardware while maintaining communication transparency.
The goal is not to replace current robot control frameworks mentioned above. The goal is to provide a fault tolerant, adaptive communication middleware which is able to replace the communication layer in existing robot control software. Beside the ability to communicate transparently over hardware barriers (e.g. from CAN to Ethernet), also the time used to set up communications within these systems should significantly be lowered in order to enable researchers in the field of robotics to concentrate more on their research topics rather than communication set up.
Fault tolerance, like automatic re-routing and redundancy, is very important in any case where connection could be lost while robots are used in hazardous surroundings in tasks like in space exploration, underwater operations or Search and Rescue missions. These requirements are based on past practical experiences in robotics and the use of different robotic frameworks.