Developing decentralised systems for industry
Professor Nick Jennings and Dr Alex Rogers demonstrated decentralised systems developed for BAE Systems, Rolls Royce and Qinetiq at a conference in London this week.
Over the last five years Professor Jennings and Dr Rogers have worked with the University of Oxford to address data fusion in industry as part of the ARGUS II project. They have developed air traffic control systems which do not need a controller for BAE and decentralised engine service scheduling for airplanes and cars around the world for Rolls Royce.
'We set out to build decentralised systems which did not need any centralised controller,' said Dr Rogers. 'The advantage of this approach is that we could develop more robust systems which could behave autonomously with minimal human intervention.”
The ARGUS project, which won The Engineer Technology and Innovation Award for Large Company and University Collaboration, combines two technologies for the first time.
The Southampton team, led by Professor Jennings, is concerned with ‘agents’, computer programs that act autonomously on behalf of the humans that they represent.
‘We’re interested in systems where a number of these autonomous agents interact with each other, where they have to co-operate, negotiate or co-ordinate,’ he said.
The Oxford team, led by Professor Steve Roberts, is applying Bayesian inference to engineering and life-science problems. When information is incomplete, Bayesian techniques can help work out what are the most probable outcomes of any particular action.
Working together the teams have developed software that allows agents to communicate with each other to solve complex problems involving uncertainty.
The conference in London marked the end of ARGUS II and Oxford and Southampton have formed a strategic partnership project with BAE called Autonomous Learning Agents for Decentralised Data and Information Networks (ALADDIN) to take the research forward. Aimed primarily at developing disaster management systems, ALADDIN is extending the use of autonomous agents and Bayesian techniques for reasoning under uncertainty into areas where resources are limited and continually shifting. For more information go to www.argusiiproject.org/