A framework to configure robot ecologies : design, development and case studies

Robot ecologies are Network Robot Systems in which multiple devices cooperate to provide services to a human user. Devices can range from full-fledged robots to distributed sensors and actuators, each providing one or more elementary services. A key problem in robot ecologies is to control and configure the col- lection of services to provide added value to the user. While Artificial Intelligence techniques have been studied for this purpose, it is often useful to provide the user with the ability to summon, compose and configure services, then control the system. Thanks to the framework implemented with this thesis, it’s possible to control and configure robot ecologies easily, quickly and safely with an automatic mechanism of resources connections. Configuring results eas- ily, indeed, in the implemented graphical environment, every user can drag and drop a resource in the right input connector, very quickly, and he’s not required to know how the system works and to write any source code line or set any variable up. A robot ecology is a strongly heterogenous environment, thus an input control, based on resource type, is also provided because safely connec- tions among different devices are necessary. This is a great opportunity for new experimentations in the ecologies because all the procedures for system configuration and analysis of re- sults are considerably simplified and usability and configurability are enormously increased, so that any user has got a higher level perspec- tive in order to carry out its researches more efficiently. For those purposes different case studies are provided, showing how solving a task can automatically be done, how tasks can be solved with different configurations executed in sequence, how some configurations can be executed in parallel and how different components can be con- figured to perform a positive feedback control.