In this paper, we investigate the hypothesis that plan recognition can significantly improve the performance of a casebased reinforcement learner in an adversarial action selection task. Our environment is a simplification of an American football game. The performance task is to control the behavior of a quarterback in a pass play, where the goal is to maximize yardage gained. Plan recognition focuses on predicting the play of the defensive team. We modeled plan recognition as an unsupervised learning task, and conducted a lesion study. We found that plan recognition was accurate, and that it significantly improved performance. More generally, our studies show that plan recognition reduced the dimensionality of the state space, which allowed learning to be conducted more effectively. We describe the algorithms, explain the reasons for performance improvement, and also describe a further empirical comparison that highlights the utility of plan recognition for this task. 1. Motivation

Adaptation (online learning) by autonomous virtual characters, due to interaction with a human user in a virtual environment, is a difficult and important problem in computer animation. In this article we present a novel multi-level technique for fast character adaptation. We specifically target environments where there is a cooperative or competitive relationship between the character and the human that interacts with that character. In our technique, a distinct learning method is applied to each layer of the character’s behavioral or cognitive model. This allows us to efficiently leverage the character’s observations and experiences in each layer. This also provides a convenient temporal distinction between what observations and experiences provide pertinent lessons for each layer. Thus the character can quickly and robustly learn how to better interact with any given unique human user, relying only on observations and natural performance feedback from the environment (no explicit feedback from the human). Our technique is designed to be general, and can be easily integrated into most existing behavioral animation systems. It is also fast and memory efficient.

The ability for a given agent to adapt on-line to better interact with another agent is a difficult and important problem. This problem becomes even more difficult when the agent to interact with is a human, since humans learn quickly and behave non-deterministically. In this paper we present a novel method whereby an agent can incrementally learn to predict the actions of another agent (even a human), and thereby can learn to better interact with that agent. We take a case-based approach, where the behavior of the other agent is learned in the form of state-action pairs. We generalize these cases either through continuous k-nearest neighbor, or a modified bounded minimax search. Through our case studies, our technique is empirically shown to require little storage, learn very quickly, and be fast and robust in practice. It can accurately predict actions several steps into the future. Our case studies include interactive virtual environments involving mixtures of synthetic agents and humans, with cooperative and/or competitive relationships. Key words: autonomous agents, user modeling, agent modeling, action prediction, plan recognition. 2