A Systematic Agent Framework for Situated Autonomous Systems

We present a formal framework of an autonomous agent asa collection of coordinated control loops, with a recurringsense, plan, act cycle. Our framework manages the information flow within the partitioned structure to ensure consistency in order to direct the flow of goals and observations ina timely manner. The resulting control structure improvesscalability since many details of each controller can be encapsulated within a single control loop. This partitioned agentdesign promises a domain-independent, scalable and robustapproach for control of real-world autonomous robots operating in dynamic environments. We validate our frameworkwith experimental results from deployments in two differentreal-world domains.

Authors

• Frederic Py
• Kanna Rajan John Ryan
• Conor McGann Frederic Py

Keywords

• Robot Planning
• Cognitive robotics
• Reactive vs deliberative approaches