Resolving Conflicts in Highly Reactive Teams

In distributed cooperation frameworks for completely autonomous agents, conflicts between the involved agents can occur due to inconsistent data available to the agents. In highly dynamic domains, such as RoboCup, it is often neccessary to accept a certain level of conflicts in order to decrease reaction time of single agents, instead of relying on error-free, but extensive communication. However, this may lead to situations in which unresolved conflicts linger and cause cooperation to break down completely. In our cooperation framework, ALICA, we designed and implemented a simple but effective approach to detect these cases and resolve them quickly through a bully algorithm.