A Quick Adaptation Method for Constraint Satisfaction in a Real-time Environment

Multi-agent problem solving in a real-time environment is one of the hardest and most fascinating research area. In this paper, we propose a method for quick adaptation to changes in a problem in a real-time environment as an extension of an Asynchronous Weak-commitment (AWC) search algorithm. The basic idea of our proposing method is to limit the scope of search. Variables which are affected by changes could be modified their previously assigned values to adapt to the changes. For the adaptation to changes in a problem, we sacrifice the optimality of a solution. If an agent can increase its utility, it can simply modify its values. In the process, an agent emphasizes the local optimality without considering the global optimality. In a real-time environment, it is not practically important to globally optimize a solution. The quick adaptation to changes can be useful for such real-time problem.