For example, games in the Civilization \(^\) simulator used in , differing primarily in its focus on the predators rather than the prey, and consequently in the details of scoring games. Such adaptivity is often critical for autonomous agentsĮmbedded in games or simulators. Thus for many kinds of task an ATA could be successful even if there are fewer agents than the number of roles demanded by the task. If necessary, an agent can alternate between roles continuously in order to ensure that sufficient progress is made on all sub-tasks. An ATA is robust because there are no critical task specialists that cannot be replaced by other members of the team it is flexible because individual agents can switch roles whenever they observe that a sub-task is not receiving sufficient attention. Intent, and either the reality or appearance of intent may need to be instilled into the individual agents in order to achieve that. That is, the team must pursue its owner’s The owner of the team, whether in the context of robotics, simulation, or games, must also be able to trust the team as a whole to work out an effective division of labor in order to get the team’s overall task done thoroughly and efficiently. Agents cannot select appropriate sub-tasks without some sort of assurance – possibly supported by observation – that the other members of the team are also selecting contextually appropriate sub-tasks. Within the team, individual agents must trust all the others to “do the right thing”. It changes the division dynamically as conditions change, and if composed of autonomous agents it must be able to organize the necessary divisions of labor without direction from a human operator. An ATA is a homogeneous team that self-organizes a division of labor in situ so that it behaves as if it were a heterogeneous team. We call such a multi-agent architecture an Adaptive Team of Agents (ATA) Inflexible, because if a client requested a 20% speed-up for the task you would not be able to simply send in 20% more robots you would have to add \(\lceil 20 \% \rceil \) more robots for each sub-task specialization, four more robots in all rather than two.Īn alternative approach is to use a team of homogeneous agents, each capable of adopting any role required by the team’s task, and capable of switching roles to optimize the team’s performance in its current context. Of a single spraying robot would reduce the entire team to half speed at the cleaning task, or the loss of the pumper robot would cause the team to fail entirely. If the individual robots were programmed or trained as sub-task specialists the team would be brittle and lacking in flexibility. Moreover, when the agents in a team are programmed or trained to optimize a pre-specified division of labor, the team may perform inefficiently if the size of the team changes – for example, if more agents are added to speed up the task – or if the scope of the task changes.įor example, suppose you owned a team of ten reactor cleaning robots, and the optimal division of labor for the cleaning task required two sprayers, seven scrubbers, and one pumper (Fig. However, heterogeneous teams of sub-task specialists are brittle: if one specialist fails then the whole team may failĪt its task. Heterogeneous teams are often used for complex tasks because they allow agents to be specialized for sub-tasks (e.g. ). The agents that comprise a multi-agent system can be either homogeneous Is a suitable platform for research into multi-agent systems as well. Have previously been identified as a possible “killer application” for artificial intelligence , and a game involving multiple autonomous agents Multi-agent systems are often formalized for entertainment as well, with instances ranging from team sports to computer games. Parties cooperating or competing at some task. Each of these domains consists of multiple autonomous Are a commonplace in social, political, and economic enterprises.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |