Sriram Gopalakrishnan, Ph.D.

Sriram Gopalakrishnan, Ph.D.

AI Research Lead at J.P. Morgan Chase | Sequential-Decision Making Problems, Human-AI Interaction, & Generative AI

DARPA SAIL-ON: Open-World Novelty Monopoly Agent

Published:

Pillar: Human-AI Interaction
DARPA SAIL-ON Novelty Handling Open-World Planning MCTS

Overview

In the DARPA SAIL-ON (Science of Artificial Intelligence and Learning for Open-world Novelty) program, autonomous agents must recognize and adapt when environment rules abruptly change without prior notice.

Our system:

  • Combined symbolic planning, execution monitoring, and Monte Carlo Tree Search (MCTS) to detect anomalous state transitions in Monopoly.
  • Dynamically revised the underlying domain transition graph to recover optimal winning policies.
  • Recognized as the Top Performing Agent in the entire DARPA SAIL-ON Monopoly domain track.