Agent Technologies for Space Applications: The DAFA Experience Francesco Amigoni, Andrea Brambilla, Michèle Lavagna Politecnico di Milano Piazza Leonardo da Vinci 32, 20133 Milano, Italy francesco.amigoni@polimi.it Rick Blake, Ian le Duc, Jonathan Page, Oliver Page SciSys UK Ltd Methuen Park, Chippenham, Wiltshire SN14 0GB, UK Sonia de la Rosa Steinz, Robin Steel VEGA Europaplatz 5, 64293 Darmstadt, Germany Quirien Wijnands ESA/ESTEC Keplerlaan 1, Noordwijk, The Netherlands Abstract — The Distributed Agents For Autonomy (DAFA) study has been performed for ESA/ESTEC by SciSys UK Ltd, VEGA, and Politecnico di Milano in 2008-2009. The aim of DAFA study has been to examine the application of distributed agent technology to increase the overall performance of a space mission. It is important to note that the distribution should be seen on a system level, including both earth- and space-based agents. From an analysis of past, present, and future space missions, the ExoMars mission was identified as a mission where the application of agent technology seemed profitable. After analysis of different agent frameworks, JADE was chosen to develop a software prototype for a scenario based on an ExoMars-like mission. This demonstrator has been used to highlight the advantages of a DAFA approach for the definition of the mission architecture. Keywords – autonomous agents; ExoMars; JADE; space operations. I. INTRODUCTION The role of autonomy and distribution in the management of space systems is following a steady progression, from single, large, expensive spacecrafts with a high degree of dependence on manual, ground control to recent missions with multiple spacecrafts and a level of ground automation. Some autonomy has been implemented also in the space segment, especially in specific areas such as attitude and orbit control and fault detection, identification and recovery. It is expected that the future space missions will move into areas where significant advantages would flow from the deployment of an increased level of autonomy distributed across the space and ground segments as a whole. The aim of DAFA study has been to examine the application of distributed agent technology to increase the overall performance of a space mission. The DAFA study has been performed for ESA/ESTEC (European Space Agency, European Space Research and Technology Centre), in the framework of the General Studies Programme by SciSys UK Ltd, VEGA, and Politecnico di Milano in 2008- 2009. The study has been structured in four parts: 1. Analysis of past, present, and future mission scenarios in the context of (distributed) autonomy. 2. Analysis of agent-based programming frameworks for the definition, design, and development of agent systems. 3. Design of architectures for applying agent technologies to a set of mission scenarios. 4. Development of a demonstration prototype of the DAFA approach for a selected reference mission scenario. The outcome of the first part has been an overview of the status of both agent applications in the space domain as well as a set of use-cases where agents can contribute. Together with the design of agent architectures for a set of specific missions (part 3), an ExoMars-like mission has been selected as an interesting scenario in which agent technologies can be evaluated. This scenario has been implemented in part 4 for evaluation against a more traditional architecture. Next to this, the current availability of agent frameworks has been evaluated (part 2). As a result, JADE was identified as a suitable agent framework for implementing the DAFA demonstrator. The goals of the demonstrator were to demonstrate the use of agents and of JADE in the selected mission, to evaluate the performance of the agent-based system with respect to a more traditional system, and to evaluate the use of metrics to measure the performance of the system. These metrics would also evaluate typical agent behaviors (e.g., chattering). This paper, applicative in its nature, reports some of the most significant findings emerged during the DAFA study. We mainly focus on the third and fourth parts of the study. In particular, we introduce the demonstrator and show some quantitative advantages of distribution and autonomy for a space system developed using agent technologies. These results constitute the main original contribution of this paper. II. RELATED WORKS The space community is currently involved in the development of autonomous multi-spacecraft missions like constellations (e.g., EU/ESA’s Galileo), cooperative groups 2010 IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology 978-0-7695-4191-4/10 $26.00 © 2010 IEEE DOI 10.1109/WI-IAT.2010.207 483