Monitoring talk: Distribution Planning
Tony Allard (DSTO and ANU)
ARTIFICIAL INTELLIGENCE SEMINAR Monitoring talkDATE: 2013-11-12
TIME: 15:00:00 - 15:30:00
LOCATION: RSISE Seminar Room, ground floor, building 115, cnr. North and Daley Roads, ANU
CONTACT: JavaScript must be enabled to display this email address.
ABSTRACT:
Logistics planning may be decomposed into two strongly coupled dual objectives. The first, distribution planning, entails efficient allocation of limited transport resources for the timely delivery of cargo through the supply chain. Economical inventory management requires the logistics system maintains service level, while not accumulating excessive levels of stock at any one location; thus minimising unnecessary warehousing costs. Historically, distribution planning and inventory management have generally been treated as two separate problems, each with independent areas of research. Developing automated solutions to this dual problem of logistics planning presents a significant challenge to the AI planning domain.
This research aims to investigate Reinforcement Learning, specifically Relational Reinforcement Learning, as a candidate approach for providing useful solutions to the logistics planning problem. The scope of the logistics planning problem being considered involves heterogeneous transports, multiple commodities and uncertainty in task requirements and arrival times. We also intend to compare the performance of a Relational Reinforcement Learning approach with other well-known planners such as Monte-Carlo Tree Search, Fast Forward and Multiagent Systems like MTAMDP.
BIO:
