CAREER: Plan-Based Integration of Control and Coherence in Intelligent Exploratory Environments

探索性环境——用户与物理或抽象空间的模拟进行交互以形成、追求和实现目标的虚拟世界——在从教育、培训、娱乐到社会互动的应用中取得了巨大成功。然而，这些系统的一个主要限制是用户在其中的活动通常是过度或不足约束的。为了解决这个问题，PI将为探索性环境中的用户交互结构开发新的基于计划的模型。规划技术将用于在环境中创造新的活动，包括系统控制的角色、环境和用户的行动。PI将以他之前在计划生成和计划相关通信方面的工作为基础，开发一种在智能探索环境中创建、监控和控制交互的体系结构。该架构将集成：规划算法，为虚拟世界中的交互创建计划，其结构易于用户理解；有效传达当前展开的基于计划的活动的话语生成算法；执行监控算法在用户界面和虚拟世界之间调解操作的执行，当用户打算执行的操作偏离系统当前的计划结构时，以适当的方式进行干预。如果成功，这项工作将大大提高我们生产引人入胜、新颖有效的虚拟环境的能力，在这种环境中，用户的领域和任务导向的学习将在一系列应用程序和环境中得到增强。

Exploratory environments - virtual worlds in which users interact with a simulation of a physical or abstract space to form, pursue and achieve their goals - have shown great success in applications ranging from education and training to entertainment to social interaction. A principal limitation of these systems, however, is that users' activities within them are typically greatly over- or under-constrained. To address this problem, the PI will develop new plan-based models for the structure of user interactions within exploratory environments. Planning techniques will be used to create novel activity within the environment that encompasses both the system-controlled characters, the environment and the actions of the user. The PI will build on his prior work in plan generation and plan-related communication to develop an architecture for creating, monitoring and controlling interaction in intelligent exploratory environments. This architecture will integrate: planning algorithms that create plans for interaction in virtual worlds whose structure is readily understandable by users; discourse generation algorithms that effectively communicate the currently unfolding plan-based activity; and execution monitoring algorithms that mediate the execution of actions between the user interface and the virtual world, intervening in appropriate ways when actions that the user intends to perform deviate from the system's current plan structure. If successful, the work will greatly increase our ability to produce engaging, novel and effective virtual environments in which users' domain- and task-oriented learning are enhanced across a range of applications and contexts.