Cooperating with a markovian ad hoc teammate
نویسندگان
چکیده
This paper focuses on learning in the presence of a Markovian teammate in Ad hoc teams. A Markovian teammate’s policy is a function of a set of discrete feature values derived from the joint history of interaction, where the feature values transition in a Markovian fashion on each time step. We introduce a novel algorithm “Learning to Cooperate with a Markovian teammate”, or Lcm, that converges to optimal cooperation with any Markovian teammate, and achieves safety with any arbitrary teammate. The novel aspect of Lcm is the manner in which it satisfies the above two goals via efficient exploration and exploitation. The main contribution of this paper is a full specification and a detailed analysis of Lcm’s theoretical properties. 1
منابع مشابه
Learning Teammate Models for Ad Hoc Teamwork
Robust autonomous agents should be able to cooperate with new teammates effectively by employing ad hoc teamwork. Reasoning about ad hoc teamwork allows agents to perform joint tasks while cooperating with a variety of teammates. As the teammates may not share a communication or coordination algorithm, the ad hoc team agent adapts to its teammates just by observing them. Whereas most past work ...
متن کاملTeamwork with Limited Knowledge of Teammates
While great strides have been made in multiagent teamwork, existing approaches typically assume extensive information exists about teammates and how to coordinate actions. This paper addresses how robust teamwork can still be created even if limited or no information exists about a specific group of teammates, as in the ad hoc teamwork scenario. The main contribution of this paper is the first ...
متن کاملTeaching and leading an ad hoc teammate: Collaboration without pre-coordination
As autonomous agents proliferate in the real world, both in software and robotic settings, they will increasingly need to band together for cooperative activities with previously unfamiliar teammates. In such ad hoc team settings, team strategies cannot be developed a priori. Rather, an agent must be prepared to cooperate with many types of teammates: it must collaborate without pre-coordinatio...
متن کاملPolicy Communication for Coordination with Unknown Teammates
Within multiagent teams research, existing approaches commonly assume agents have perfect knowledge regarding the decision process guiding their teammates’ actions. More recently, ad hoc teamwork was introduced to address situations where an agent must coordinate with a variety of potential teammates, including teammates with unknown behavior. This paper examines the communication of intentions...
متن کاملHoc Teamwork by Learning Teammates ’ Task ( JAAMAS Extended
We address ad hoc teamwork, where an agent must coordinate with other agents in an unknown common task without pre-defined coordination. We formalize the ad hoc teamwork problem as a sequential decision problem and propose (i) the use of an online learning approach that considers the different tasks depending on their ability to predict the behavior of the teammate; and (ii) a decision-theoreti...
متن کامل