A web service-based approach for developing self-adaptive systems

The current development on self-adaptive systems mainly focuses on autonomy and self-containment where decision-making solely depends on the local knowledge base within the system. This limits the evolution of the knowledge base for making more precise decisions. There have been some recent works using the cloud for the knowledge base. However, they suffer from the overhead caused by the communication with the cloud. In this work, we propose a hybrid approach for developing self-adaptive systems using both the local knowledge base in the vehicle and the global knowledge base provided via a web service. The global knowledge base is shared and evolves by multiple vehicles through the web service. We validate the approach using Gazebo, a 3D simulation environment for robotic systems. The results show 96% precision in identifying objects with a viable overhead introduced by the web service and 40% improvement in precision over the traditional approach.