Resilient Reasoning Robotic Co-operating Systems (R3-COP)
Autonomous systems are one of the important instantiations of embedded systems in middle-range future, simply because the application domains are extremely manifold, from household to transportation and manufacturing, etc. Consequently, Autonomous Systems were identified as one of the targets of the ARTEMIS SRA.
Today a large number of different approaches and platforms exist, rendering an economic realisation of such systems currently inefficient (besides the manufacturing domain, where robots are already industrially exploited). Simultaneously, as such systems increasingly share environment and even closely cooperate with humans, there is an urgent need for providing any possible means and measures to assert and guarantee the dependability (in particular safety and robustness) of intelligent autonomuos systems. The project aims at the elaboration of intelligent computing platform and such testing and verification methods that can be effectively used in case of complex autonomous systems.
The contributions of BME in the project focus on the following areas:
1. Research and development in embedded intelligence: The principal role of learning in autonomus systems is to elevate the intelligence level and the robustness against phenomena not anticipated fully in design time. The research in machine learning will extend the expert knowledge addressing the possibilities and limits of various approaches and tools. Machine learning solutions will be seamlessly plugged into the components of the working system.
2. Scenario-based testing of autonomous systems: The objective for developing a scenario based testing approach is to provide algorithms and tools for automated testing of cooperation and autonomous behaviour. To achieve this aim, new coverage metrics and a scenario language will be developed that support the representation of context changes, adaptability of behaviour and cooperation. Robustness test cases will be generated automatically using the scenarios and taking into account fault patterns, stressful context and extreme behaviour.
