Developing BDI-Based Robotic Systems with ROS2.

Robots are very effective in automatizing repetitive operations leveraging hard-coded control, actuation, and sensing algorithms. Current industrial automation trends demand combining low-level reactive primitives with high-level autonomy capabilities (e.g., reasoning and planning). Recent robotic reactive architectures provide capabilities to reliably sense the environment and promptly react to stimuli, but their autonomy capabilities are in an early stage and present several limitations. The BDI model has been proposed as a reference model to build autonomous agents, but it does not provide any kind of conceptual and developing framework to connect the reasoning and planning capabilities with the lower level reactive functionalities of a robotic system. In this paper, we propose an architecture supporting BDI-based solutions to develop agents with deliberation and priority aware executions of plans, and that consider deadline-aware prioritization of desires. We build our architecture on top of ROS 2 (a standard robotic framework) leveraging and extending state-of-the-art ROS planning infrastructures. We provide a novel development tool-kit that allows for the implementation of autonomous robotic systems. Finally, we show by means of a realistic industrially inspired scenario how to use the developed tool-kit.