DRESS-ML: A Domain-specific Language for Modelling Exceptional Scenarios and Self-adaptive Behaviours for Drone-based Applications

Drones are gaining attention due to its possibility to support wide different types of applications. Since they can operate in different environments, it is possible to encounter uncertainties and exceptional situations, not initially predicted, during the use of drone-based applications. In this realm, self-adaptive strategies have been successfully used to guarantee resilience and continuous execution of such applications despite environment changes. Although some modelling approaches emerged to represent drone concepts, they are limited to model only expected flight plans or include few environmental conditions and drone resources, which restrict considerably their use. To mitigate those problems, this work proposes a domain-specific language, called DRESS-ML, which allows modelling exceptional situations and self-adaptive behaviours for drone-based applications. It relies on the Given-When-Then template used in the Behaviour-driven development (BDD) technique and the some of the main Aspect-oriented Programming concepts. We validate the applicability of our language through a proof of concept regarding an example application that uses a drone to monitor a forest to search for fire spots. Drones are gaining attention due to their possibility of supporting diverse applications and environments, such as search-and-rescue, surveillance, and goods delivery. Due to this variety of applications, drones can face both uncertainties and exceptional situations that they did not initially foresee during flight plan. In this sense, providing the ability to monitor the system and its environment and to change the system to ensure resilience and continuous execution are benefits of self-adaptation strategies. Although some studies have proposed approaches to model drone concepts, they are limited to model only expected flight plans or include few environmental conditions and drone resources, which restrict considerably their use. This work proposes a domain-specific language, called DRESS-ML, which allows modelling exceptional scenarios and self-adaptive behaviors to mitigate those problems. The language is based on an well known structure used for specifying system behaviour and the main concepts of a programming paradigm that injects encapsulated behaviours in the system. A practical example that uses a drone to monitor a forest to search for fire spots is used to evaluate the proposed language, demonstrating its applicability to model various exceptional scenarios.