Information Technology at Notre Dame

Intelligent Systems

Information Technology has long since grown from simple rote processing of reams of data at extremely high speeds. Today, observations are converted from raw measurements into "information" about the environment, processed "as information," and then used to effect how a system interacts with the outside world to achieve some "goal." Such terms as "artificial intelligence," "expert systems," "fuzzy logic," "actors", and "neural nets" are common in the technology, but a bit esoteric in terms of their perceived applicability. ND has been making great strides, as indicated below.

CCPWNS

The Computer Controlled Power Wheelchair Navigation System (CCPWNS) project Headed by Prof. Steven Skaar in the AME department provides navigational autonomy for disabled users who, due to a mix of disabilities, which may include blindness, cannot control their own power wheelchairs within their home or workplace setting. Based upon estimation algorithms that apply a mix of wheel odometry and visual detection of prepositioned wall cues, the system constructs a feasible connection between the current chair position and a user-input destination by piecing together previously taught path segments. The performance is precise and robust. Ultrasound is used to avoid collision with new obstacles (whereupon the current path is executed in reverse); and a voice-based user interface is demonstrated.
The working chair was displayed at the Indiana State Fair in August of 2002, and it was also demonstrated in the spring of 2001 at a Dept. of Veterans Affairs Forum at the Hines VA Hospital, Hines, IL, where the poster discussing the project won first place.

Thermal Engineering

Prof. Mihir Sen of AME has applied artificial neural networks and fuzzy logic modeling to problems in the thermal engineering area. Specific examples include the performance prediction of heat exchangers and flow wakes as well as dynamic control of temperature.

Artificial Intelligence and Robotics Laboratory

The Artificial Intelligence and Robotics Laboratory, founded by Professor Matthias Scheutz, continues a tradition of research in artificial intelligence that started at Notre Dame as early as 1962. (under the guidance of Professor Kenneth Sayre, who directed a group of researchers under NSF sponsorship at the "Philosophic Institute").  While the efforts then were mainly concentrated on pattern recognition, the current research focuses on the design of software architectures for simulated and robotic agents. A thorough analysis of different types of agents, the architectures they require, and their capabilities is the prerequisite for understanding their potential applications in application domains as diverse as personal assistants, believable characters for the entertainment industry, E-commerce, automated knowledge processing, and many others.

  • Affective Agent Architectures. Work in this area involves conceptual analyses of affect concepts in terms of functional components of agent architectures and the kinds of processes they can bring about as well as comparisons of different kinds of agent architectures for affective agents, their components and functional capacities.
  • Complex Complete Robotic Agents. The focus here lies on designing architectures for complete robotic agents (such as "robotic waiters" or "soccer playing robots"). Architectures are implemented and tested on various robots (e.g., the ActivMedia PeopleBot and Pioneer 2DXE robots) using the AGES agent architecture development and test environment as well as the integrated neural network simulator NNSIM, both under development in the AIROLAB. AGES permits users to define agent architectures in a generic way (at several levels of abstraction) and can run the architectures without modification on simulated and real robots.
  • Affective Agent Control. Here, the general trade-offs of agents controlled by affective states as compared to agent controlled by more classical deliberative methods are investigated using simulation experiments in the artificial life simulator SIMWORLD. SIMWORLD is based on the free SIMAGENT toolkit (developed by Aaron Sloman), which provides functionality for running different interacting agents and objects in a simulated, continuous environment. The agents are controlled by rules written in the powerful rule interpreter POPRULEBASE (which is part of SIMAGENT). New behaviors of agents can be defined without any knowledge of POP11 the underlying programming language of S IMAGENT and SIMWORLD). SIMWORLD is used together with a distributed simulation experiment server, which dynamically schedules and supervises experiments on remote hosts. Once the experiments have finished, the server gathers statistics and produces a summary report in several formats (e.g., html or latex).