SUMMARY
Ad hoc teamwork refers to the challenge of designing agents that can influence the behavior of a team, without prior coordination with its teammates. This paper considers influencing a flock of simple robotic agents to adopt a desired behavior within the context of ad hoc teamwork. Specifically, we examine how the ad hoc agents should behave in order to orient a flock towards a target heading as quickly as possible when given knowledge of, but no direct control over, the behavior of the flock. We introduce three algorithms which the ad hoc agents can use to influence the flock, and we examine the relative importance of coordinating the ad hoc agents versus planning farther ahead when given fixed computational resources. We present detailed experimental results for each of these algorithms, concluding that in this setting, inter-agent coordination and deeper lookahead planning are no more beneficial than short-term lookahead planning.