A Globally Optimal Algorithm for TTD-MDPs

In this paper, we discuss the use of Targeted Trajectory Distribution Markov Decision Processes (TTD-MDPs)–a variant of MDPs in which the goal is to realize a specified distribution of trajectories through a state space–as a general agent-coordination framework.

Authors

• Sooraj Bhat Charles L. Isbell Jr.
• David L. Roberts North Carolina State University
• Mark J. Nelson Michael Mateas
• Charles L. Isbell Jr.
• Michael Mateas University of California, Santa Cruz

Keywords

• Markov decision processes
• interactive entertainment
• convex optimization