Stochastic Scheduling and Planning Using Reinforcement Learning
Abstract
This project investigated the extension of reinforcement learning (RL) methods to large-scale optimization problems relevant to Air Force operations planning, scheduling, and maintenance. The objectives of this project were to: (1) investigate the utility of RL on large-scale logistics problems; (2) extend existing RL theory and practice to enhance the ease of application and the performance of RL on these problems; and (3) explore new problem formulations in order to take maximal advantage of RL methods. A method using RL to modify local search cost functions was developed and shown to yield significant improvement over a traditional local search method on a core vehicle routing problem. A new method for stochastic dynamic optimization was studied, a theoretical result proven, and utility demonstrated using a simulated aerial mission planning task. A learning-based method for optimizing subproblem selection in divide-and-conquer approaches was developed and demonstrated on graph-coloring and on a multiple-vehicle mission planning task.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 25, 2000
- Accession Number
- ADA386083
Entities
People
- Andrew G. Barto
Organizations
- University of Massachusetts Amherst