Research Experiences for Undergraduates in Distributed Rational Agents

Open House 2003

Nine undergraduate participants of the REU summer program participated in the REU Open House 2003 to present their research projects.

• Intelligent Wireless Utility Meter

  The goal of this project was to create a system that can compute and monitor statistics related to utility usage, and exchange information wirelessly with a user. The project was built around a two-way pager module provided by Advantra Inc. One REU participant then designed and implemented a set of software and interface modules to provide the desired functionality. The Figure shows the hardware setup and the software design of the wireless utility metering and monitoring system. All the components in the student's design are linked using CORBA to make it possible to locate different components of the system on separate machines. For example, the graphical user interface can be located on a different computer than the statistics module which monitors the utility meter directly and computes measures of the usage patterns within the home. These statistics are then used by the paging module to answer information requests from the user and to send alarms if the current usage crosses some pre-set thresholds, indicating the occurrence of an unusual event. All alarm parameters and requests can be issued either from a wireless pager or using e-mail from a computer, and the requested information and alarms are sent through the pager network to either a pager or an e-mail account.
  

• Automated Multimedia Device Control

  This project was intended to add additional devices to the suite of components that can be controlled automatically in MavHome. In particular, it was aimed at providing computer access and intelligent control for entertainment devices such as TVs, VCRs, or stereo systems using a SmartLINCÔ computer/infrared interface. To achieve this, the student had to make herself familiar with serial protocols and develop the interface component. This was then augmented with additional control commands that permit the programming of functions such as sleep timers for systems that do not natively support them. This interface is implemented as a CORBA server and paired up with a graphical remote control GUI that permits the user to operate the entertainment device from any computer within the home.

• Connection Hardware for Intelligent Sensor Networks

  One of the challenges in smart homes is that a potentially large number of sensors are required to provide information about the state of the home. It is therefore essential that a reliable sensor network can be established that provides access to the data without overwhelming the network bandwidth. For this purpose, one REU student with the help of an additional undergraduate student set out to design and build hardware and embedded software for smart sensor network modules. The goal here was to construct a system that permits large numbers of sensors to be connected and that can make intelligent decisions about the set of data readings that should be transmitted. To do this the students studied different technologies and came up with a master-slave design that uses PIC microprocessors on slave modules to read sensors and make decisions about the transmission of data. Each module can be connected to other modules using an I2C serial bus. The serial bus, in turn, is connected to a master board that connects to Ethernet. This system permits flexible and fast reconfiguration by adding and removing slave modules and provides a relatively low-cost solution that is highly reliable. By the end of the summer, the students built and tested a prototype board.

• Automatic Service Discovery

  In the system underlying the MavHome smart home project, a set of rational agents monitor inhabitants and make decisions about what devices to automate. As a result, a larger number of computers and processes have to interact and use each other's services. To make such a distributed system robust it is important that newly started programs can readily find the available services in the home and interface with them. While a number of technologies for this exist, each has drawbacks. The goal of this project was to research and develop a methodology to augment CORBA servers and clients with the capability of automatic discovery without the need for explicit knowledge of the IOR identifier for the server's location. To achieve this, one REU student researched different service discovery technologies and designed and successfully implemented a system that combines CORBA and Rendezvous (ZERO configuration) to facilitate the automatic distribution of IOR strings to new CORBA clients. The result is a system where new CORBA agents can automatically discover all available services and connect to them without knowledge of their physical location.

• Human/Computer Interfaces for Smart Home Environments

  Another important aspect of a smart home is the interface it provides to the inhabitants in order to permit them to efficiently monitor the state of the home and to interact with it in an intuitive manner. Over the course of the summer, two REU students investigated different techniques and approaches to human/computer interfaces and developed graphical interfaces for personal computers and for portable devices based on a common data layer. To permit flexibility the final design used CORBA to communicate with the devices in the home and an XML packet server to distribute the information about the state of the home to the graphical interfaces. As a result of this design, multiple graphical interfaces could be active simultaneously. The Figure shows a block diagram of the complete architecture and a screen shot of the graphical interface on a PDA. A second interface for personal computers was also developed and both graphical interfaces were used to successfully monitor and control the devices in the laboratory.
  

• Motion Detection Using Computer Vision

  The goal of this project was to implement a motion detection and tracking system for a camera network in the home. Starting from a motion detection package developed at UTA, an REU student implemented this system by expanding the software to a scenario where multiple cameras are available. To achieve this the student had to research different motion tracking techniques and become familiar with digital image processing. The resulting system was capable of finding moving targets in all the cameras and of providing information about the locations of the moving objects.

• Smart Kitchen

  One important part of a home that lends itself to smart technologies is the kitchen. In this project one REU student developed and implemented a kitchen system equipped with a smart shelf, a countertop projection display, and an interface consisting of a gesture pad. In collaboration with the students involved in the sensor network development, the student constructed a sensory mechanism for a kitchen shelf that tracks the state of the cabinet doors as well as the weight distribution on the shelf. As a result, the system can determine when items are removed or added. To determine the types of items on the shelf, the student also installed an RF ID tag reader that was donated to the research project by Texas Instruments. However, since the reader only arrived at the end of the summer, it could not be fully integrated with the shelf system within the course of the REU program.

  In addition to the shelf system, the REU student's research also included a countertop information system for the kitchen. Here, the student developed an implemented a system that uses a projector to display important information, including recipes and ingredient lists, directly onto the kitchen counter. Using a gesture pad, the user can easily navigate through the different pieces of information without the need for a keyboard.

For additional information regarding this program please contact:

Manfred Huber
University of Texas at Arlington
Department of Computer Science and Engineering
Box 19015
Arlington, TX 76019-0015
Email: reu@cse.uta.edu

* This project is funded by NSF grant EIA-0139564