User-Generated AI for Interactive Digital Entertainment

User-generated content is everywhere: photos, videos, news, blogs, art, music, and every other type of digital media on the Social Web. Games are no exception. From strategy games to immersive virtual worlds, game players are increasingly engaged in creating and sharing nearly all aspects of the gaming experience: maps, quests, artifacts, avatars, clothing, even games themselves. Yet, there is one aspect of computer games that is not created and shared by game players: the AI. Building sophisticated personalities, behaviors, and strategies requires expertise in both AI and programming, and remains outside the purview of the end user.

To understand why Game AI is hard, we need to understand how it works. AI can take digital entertainment beyond scripted interactions into the arena of truly interactive systems that are responsive, adaptive, and intelligent. I discuss examples of AI techniques for character-level AI (in embedded NPCs, for example) and game-level AI (in the drama manager, for example). These types of AI enhance the player experience in different ways. The techniques are complicated and are usually implemented by expert game designers.

I argue that User-Generated AI is the next big frontier in the rapidly growing Social Gaming area. From Sims to Risk to World of Warcraft, end users want to create, modify, and share not only the appearance but the “minds” of their characters. I present my recent research on intelligent technologies to assist Game AI authors, and show the first Web 2.0 application that allows average users to create AIs and challenge their friends to play them—without programming. I conclude with some thoughts about the future of AI-based Interactive Digital Entertainment.

CMU Robotics & Intelligence Seminar, September 28, 2009
Carnegie-Mellon University, Pittsburgh, PA.

Drama Management and Player Modeling for Interactive Fiction Games

A growing research community is working towards employing drama management components in story-based games. These components gently guide the story towards a narrative arc that improves the player’s gaming experience. In this paper we evaluate a novel drama management approach deployed in an interactive fiction game called Anchorhead. This approach uses player’s feedback as the basis for guiding the personalization of the interaction.

The results indicate that adding our Case-based Drama manaGer (C-DraGer) to the game guides the players through the interaction and provides a better overall player experience. Unlike previous approaches to drama management, this paper focuses on exhibiting the success of our approach by evaluating results using human players in a real game implementation. Based on this work, we report several insights on drama management which were possible only due to an evaluation with real players.

Read the paper:

Drama Management and Player Modeling for Interactive Fiction Games

by Manu Sharma, Santi Ontañón, Manish Mehta, Ashwin Ram

To appear in Computational Intelligence journal

www.cc.gatech.edu/faculty/ashwin/papers/er-09-10.pdf

Run-Time Behavior Adaptation for Real-Time Interactive Games

Intelligent agents working in real-time domains need to adapt to changing circumstance so that they can improve their performance and avoid their mistakes. AI agents designed for interactive games, however, typically lack this ability. Game agents are traditionally implemented using static, hand-authored behaviors or scripts that are brittle to changing world dynamics and cause a break in player experience when they repeatedly fail. Furthermore, their static nature causes a lot of effort for the game designers as they have to think of all imaginable circumstances that can be encountered by the agent. The problem is exacerbated as state-of-the-art computer games have huge decision spaces, interactive user input, and real-time performance that make the problem of creating AI approaches for these domains harder.

In this paper we address the issue of non-adaptivity of game playing agents in complex real-time domains. The agents carry out run-time adaptation of their behavior sets by monitoring and reasoning about their behavior execution to dynamically carry out revisions on the behaviors. The behavior adaptation approaches has been instantiated in two real-time interactive game domains. The evaluation results shows that the agents in the two domains successfully adapt themselves by revising their behavior sets appropriately.

Read the paper:

Run-Time Behavior Adaptation for Real-Time Interactive Games

by Manish Mehta, Ashwin Ram

To appear in IEEE Transactions on Computational Intelligence and AI in Games

www.cc.gatech.edu/faculty/ashwin/papers/er-09-09.pdf

Using Meta-Reasoning to Improve the Performance of Case-Based Planning

Case-based planning (CBP) systems are based on the idea of reusing past successful plans for solving new problems. Previous research has shown the ability of meta-reasoning approaches to improve the performance of CBP systems. In this paper we present a new meta-reasoning approach for autonomously improving the performance of CBP systems that operate in real-time domains.

Our approach uses failure patterns to detect anomalous behaviors, and it can learn from experience which of the failures detected are important enough to be fixed. Finally, our meta-reasoning approach can exploit both successful and failed executions for meta-reasoning.

We illustrate its benefits with experimental results from a system implementing our approach called Meta-Darmok in a real-time strategy game. The evaluation of Meta-Darmok shows that the system successfully adapts itself and its performance improves through appropriate revision of the case base.

Read the paper:

Using Meta-Reasoning to Improve the Performance of Case-Based Planning

by Manish Mehta, Santi Ontañón, Ashwin Ram

International Conference on Case-Based Reasoning (ICCBR-09), Seattle, July 2009

www.cc.gatech.edu/faculty/ashwin/papers/er-09-06.pdf

Collaborative Information Access: A Conversational Search Approach

Knowledge and user-generated content is proliferating on the web in scientific publications, information portals and online social media. This knowledge explosion has continued to outpace technological innovation in efficient information access technologies. In this paper, we describe methods and technologies for “Conversational Search” as an innovative solution to facilitate easier information access and reduce the information overload for users.

Conversational Search is an interactive and collaborative information finding interaction. The participants in this interaction engage in social conversations aided with an intelligent information agent (Cobot) that provides contextually relevant search recommendations. The collaborative and conversational search activity helps users make faster and more informed search and discovery. It also helps the agent learn about conversations with interactions and social feedback to make better recommendations. Conversational search leverages the social discovery process by integrating web information retrieval along with the social interactions.

Read the paper:

Collaborative Information Access: A Conversational Search Approach

by Saurav Sahay, Anu Venkatesh, Ashwin Ram

ICCBR-09 Workshop on Reasoning from Experiences on the Web (WebCBR-09), Seattle, July 2009

www.cc.gatech.edu/faculty/ashwin/papers/er-09-05.pdf

Authoring Behaviors for Games using Learning from Demonstration

Behavior authoring for computer games involves writing behaviors in a programming language. This method is cumbersome and requires a lot of programming effort to author the behavior sets. Further, this approach restricts the behavior set authoring to people who are experts in programming.

This paper describes our approach to design a system that allows a user to demonstrate behaviors to the system, which the system uses to learn behavior sets for a game domain. With learning from demonstration, we aim at removing the requirement that the user has to be an expert in programming, and only require him to be an expert in the game. The approach has been integrated in a easy-to-use visual interface and instantiated for two domains, a real-time strategy game and an interactive drama.

Read the paper:

Authoring Behaviors for Games using Learning from Demonstration

by Manish Mehta, Santiago Ontañón, Tom Amundsen, Ashwin Ram

ICCBR-09 Workshop on Case-Based Reasoning for Computer Games, Seattle, July 2009

www.cc.gatech.edu/faculty/ashwin/papers/er-09-07.pdf

An Ensemble Learning and Problem Solving Architecture for Airspace Management

In this paper we describe the application of a novel learning and problem solving architecture to the domain of airspace management, where multiple requests for the use of airspace need to be reconciled and managed automatically. The key feature of our “Generalized Integrated Learning Architecture” (GILA) is a set of integrated learning and reasoning (ILR) systems coordinated by a central meta-reasoning executive (MRE). Each ILR learns independently from the same training example and contributes to problem-solving in concert with other ILRs as directed by the MRE. Formal evaluations show that our system performs as well as or better than humans after learning from the same training data. Further, GILA outperforms any individual ILR run in isolation, thus demonstrating the power of the ensemble architecture for learning and problem solving.

Read the paper:

An Ensemble Learning and Problem Solving Architecture for Airspace Management

by XS Zhang et al.

International Conference on Innovative Applications of Artificial Intelligence (IAAI-09), Pasadena, CA, July 2009

www.cc.gatech.edu/faculty/ashwin/papers/er-09-03.pdf

Goal-Driven Learning in the GILA Integrated Intelligence Architecture

Goal Driven Learning (GDL) focuses on systems that determine by themselves what has to be learned and how to learn it. Typically GDL systems use meta-reasoning capabilities over a base reasoner, identifying learning goals and devising strategies. In this paper we present a novel GDL technique to deal with complex AI systems where the meta-reasoning module has to analyze the reasoning trace of multiple components with potentially different learning paradigms. Our approach works by distributing the generation of learning strategies among the different modules instead of centralizing it in the meta-reasoner. We implemented our technique in the GILA system, that works in the airspace task orders domain, showing an increase in performance.

Read the paper:

Goal-Driven Learning in the GILA Integrated Intelligence Architecture

by Jai Radhakrishnan, Santi Ontañón, Ashwin Ram

International Joint Conference on Artificial Intelligence (IJCAI-09), Pasadena, CA, July 2009

www.cc.gatech.edu/faculty/ashwin/papers/er-09-02.pdf

Learning from Human Demonstrations for Real-Time Case-Based Planning

One of the main bottlenecks in deploying case-based planning systems is authoring the case-base of plans. In this paper we present a collection of algorithms that can be used to automatically learn plans from human demonstrations. Our algorithms are based on the basic idea of a plan dependency graph, which is a graph that captures the dependencies among actions in a plan. Such algorithms are implemented in a system called Darmok 2 (D2), a case-based planning system capable of general game playing with a focus on real-time strategy (RTS) games. We evaluate D2 with a collection of three different games with promising results.

Read the paper:

Learning from Human Demonstrations for Real-Time Case-Based Planning

by Santi Ontañón, Kane Bonnette, Praful Mahindrakar, Marco Gómez-Martin, Katie Long, Jai Radhakrishnan, Rushabh Shah, Ashwin Ram

IJCAI-09 Workshop on Learning Structural Knowledge from Observations, Pasadena, CA, July 2009

www.cc.gatech.edu/faculty/ashwin/papers/er-09-04.pdf

On Similarity Measures based on a Refinement Lattice

Retrieval of structured cases using similarity has been studied in CBR but there has been less activity on defining similarity on description logics (DL). We present an approach that allows us to present two similarity measures for feature logics, a subfamily of DLs, based on the concept of “refinement lattice”. The first one is based on computing the anti-unification (AU) of two cases to assess the amount of shared information. The second measure decomposes the cases into a set of independent “properties”, and then assesses how many of these properties are shared between the two cases. Moreover, we show that the defined measures are applicable to any representation language for which a refinement lattice can be defined. We empirically evaluate our measures comparing them to other measures in the literature in a variety of relational data sets showing very good results.

Read the paper:

On Similarity Measures based on a Refinement Lattice

by Santi Ontañón and Enric Plaza

in ICCBR 2009, LNAI 5650, pp 240 – 255

www.cc.gatech.edu/faculty/ashwin/papers/er-09-11.pdf