Rapid response agents

Helen Knight reports on an automated disaster response system that can assess emergencies and decide how best to deploy rescue services.

An automated disaster response system that can assess emergencies and decide how best to deploy rescue services is being developed in a research partnership between BAE Systems and Southampton University.

The system will consist of a number of autonomous software agents – adaptive systems that act on a user’s behalf – capable of obtaining information on a military or civilian incident from a variety of sources and co-operate on how best to respond.

The agents, being developed in a £3.5m research programme, could be used on the battlefield or for homeland security and will be capable of communicating and collaborating with each other, according to Prof Nick Jennings, head of the intelligence, agents and multimedia group at the university’s school of electronics and computer science, and the partnership’s academic director.

The system will also use artificial intelligence techniques such as machine learning to allow the agents to learn to distinguish between reliable and unreliable information sources. Possible sources could include news wires, the emergency services and the army.

‘We want the agents to decide how to plan for an incident, so if there was an incident in London they might decide to send a certain number of ambulances to Kings Cross. You would typically have agents connected with certain resources, such as a fleet ofambulances from one station, or a fleet of fire engines.’ ‘We will also create agents to deal with individual aspects of a problem, so you would have an agent to deal with evacuating an area, and then another to get the emergency services to the scene. These agents would interact with each other,’ said Jennings.

But in major incidents, information is constantly being updated or changed, so the agents would be capable of altering their plans as new details emerged, he said.

‘They would start to enact the plan, but then a much bigger incident might happen, so they would need to re-plan. You might also get people at the incident, and find out things are not exactly as you first thought they were. The system will respond to what is going on in the environment.’

The agents will participate in ‘auctions’, and negotiate with each other on how best to allocate the resources, said Jennings. ‘Auctions are a simple way of allocating resources. There are a finite number of ambulances and you want to allocate them to the agents or plans that can make the best use of them. Auctions are very good at allocating services to those that need them most,’ he said.

The five-year project will involve a network of academic groups from around the world, and will also include defence and electronics systems specialist AMS, a joint venture between BAE Systems and Finmeccanica. By the end of the project the team hopes to have demonstrated how to build a full system and produced some prototypes, although any commercialisation of the project will be carried out by BAE and any other suitable partners.

As the project progresses, the researchers also hope to produce some individual technologies and techniques that can be used before the full system is completed.