by Robin Moore

Given the classic back-to-school question, “How did I spend my summer vacation?” St. Stephen’s Episcopal School senior Mark VanMiddlesworth would submit a most surprising essay. He spent a good part of the summer helping a University of Texas team develop a computer program that won the 2005 International Trading Agent/Supply Chain Management Competition held Aug. 1-3 in Edinburgh, Scotland.

Over the course of the summer, the team, led by Peter Stone, Ph.D., head of UT’s Artificial Intelligence Lab, developed the TacTex-05, an autonomous computer program that can learn, experiment, and behave independently of human control. Others working on Stone’s project included UT graduate student David Pardoe and undergraduate Jan Ulrich.

“I first learned of Dr. Stone's research when I attended a robot soccer demonstration in his lab last year,” VanMiddlesworth said. “Computer programming fascinates me. I looked at the research he was doing, and after discussing it, we decided that I could make a contribution in the Trading Agent Competition this summer.”

In the contest, agents, or autonomous computer programs designed by 30 university teams from around the world, competed against one another in a game simulating the manufacture of personal computers. Agents managed parts inventories, competed for customers, negotiated with suppliers, and performed a range of other decision-making tasks over a virtual 220-day period. The agent who finished the game with the most money in the bank was declared winner.

TacTex-05 took first place in the virtual manufacturing competition, defeating agents designed by teams from Harvard, Carnegie-Mellon, McGill, Peking University, and 26 other top research institutions.

“Going into the final rounds, we had no idea whether we were going to win,” VanMiddlesworth recalled. “The other agents had beaten us earlier, and the improvements we had made the night before had not yet been tested in a game against such stiff competition. We were all surprised at how well we performed in the last day of the competition. It was great to feel like I had played a significant role in something so important.”

In the growing global economy, autonomous agents like TacTex-05 could have tremendous impacts on supply-chain manufacturing processes, such as automobile assembly lines. Today most supply chains are managed by humans, who must coordinate multiple manufacturers and parts suppliers and make speculative decisions based on quickly changing markets.

“Autonomous agents may be able to make much quicker, more flexible decisions,” said Stone, an assistant professor of computer sciences. “Tac Tex-05 and similar programs could make the entire supply chain more efficient and responsive to environmental changes.”

VanMiddlesworth, who is considering a career in software development, is grateful for the opportunity to be on the team this summer and build his programming skills.

“Dr. Stone, a leader in the artificial intelligence community, is brilliant, as are David and Jan,” he said. “I learned a lot by being able to locate problems and suggest fixes along with the rest of the team. I took control of one aspect of our agent, and it was my responsibility to work on that part of our strategy. What I worked on became a significant part of our agent, and it was great to feel like I really made a difference.”