Abstract-The ATLAS experiment at CERN will require a large amount of computing resources for the online analysis system. The software and communication protocols in the ATLAS Online analysis system are optimized for a cluster environment. We setup a geographically distributed testbed to evaluate the implications of integrating remote computing resources in this environment. This paper reports on the integration scenarios and analyzes the achieved performance. We highlight limitations in the communication protocols and suggest solutions for solving them. A proposal for employing Gridenabled resources to allow for on-demand expansion of the computing capabilities is presented at the end of the paper.

Robotics systems in space must deal with a host of software challenges in addition to and amplified by the challenges of a hostile space environment and the remoteness in which they operate. Software automation has been used to good effect in managing some of these challenges to control single robotic systems, even multi-functional systems, in previous unmanned missions without the physical presence of human operators. However, these previous control software artifacts tend to be platform-specific, single-systemfocused and not really applicable to multi-agent teams of multi-functional, multi-generational robotic systems. In this paper, we discuss a software architecture based on the Multi-Agent Distributed Adaptive Resource Allocation 1 (MADARA) and Group Autonomy for Mobile Systems 2 (GAMS) open source middleware projects that is intended to be deployed in a multi-agent, multi-functional robotic system called the Keck Institute for Space Studies Multi-Planetary Smart Tile. We discuss our solution approaches to addressing scalability and quality-of-service in deployments of multi-agent systems, codifying group intelligence in hostile space environments, portability for future missions and systems, and assurance and verification of software controllers and algorithms.