Tizar Rizano, University of Trento, Italy
Luigi Palopoli, University of Trento, Italy
Luca Abeni, University of Trento, Italy

In the DALi project, we pursue a mobility aid device that supports navigation in crowded and unstructured spaces by acquiring sensory information by anticipating the intent of human agents and by deciding the path that minimizes the risk of accidents. The cognitive walker (c-walker) is designed to assist adults with non-severe cognitive abilities in the navigation of complex environment such as shopping centers and airports. This is a paradigm of a much wider class of distributed robotic applications that are called to operate in real-time with the environment and interact with humans or with different devices.

The c-Walker integrates several modules and relies on different types of sensors that convey information on the surrounding environment. In particular, the position and the velocity of other people in the environment is detected by video sensors, while gyroscopes encoders and RFID readers are used to localize the c-Walker within a map. The same level of complexity is on the software architecture, that is comprised of modules for video analysis, mission planning, short term planning and control. These services interact with a geo-spatial database that store relevant information about the environment. The geo-spatial database maintains a consistent description of the environment, where each model inserts additional information layers.

The integration of this complex network of modules calls for a middleware solution striking a good tradeoff between conflicting needs such as: modularity, architecture independence, re-use, easy access to the limited hardware resources and real--time constraints. Several middleware architectures proposed in the last years (each one with a well maintained binding on the Linux Kernel and on the most used network protocols) offer reliable and easy to use abstractions and intuitive publish-subscribe mechanism that can simplify system development to a good degree. However, a complete compliance with the different requirements of assistive robotics application (first and foremost real-time constraints) remains to be investigated. In this paper we evaluate the performance of these solutions in terms of latency, maximum connections, and processor utilization. This comparison is used a a cornerstone for the development of a reliable software architecture for the c-Walker.