Performance of Multiple Agents in an Unknown Environment

Operating in an unknown environment is inherent to autonomous systems. This paper addresses two problems concerning efficient navigation in unknown environments. First, multiple agents can have increased performance compared to single agents. The problem of determining strategies for a cooperating group of robots in an unknown environment in order to perform a task more efficiently than a single vehicle is discussed. Second, we discuss simultaneously mapping an environment and navigating to a goal. This is a more natural strategy for an autonomous robot as irrelevant parts of the environment will not be mapped. An aspect of this particular problem is to use continuously obtained visual information and update the strategy of the robot accordingly. Two simple examples show that a game theoretic approach is a possible way of dealing with these problems. A third example uses a non-linear program to solve a similar navigation problem.