Hierarchical Planning for Agent-Based Traffic Management and Control

A planning method for traffic management and control is developed in this paper based on Hierarchical Task Network (HTN) planning technology. This method overcomes some weaknesses of goal-oriented algorithm Traffic Problem Solver (TPS) that is applied in the knowledge-based agent called Traffic Congestion Manager (TCM) to deal with real-time traffic congestion problems. Based on the domain knowledge, such as the definition of control actions, extracted from TCM, a domain model with domain description and problem description is built and tested by JSHOP2. The test result shows that, compared with TPS, the HTN planning method can provide a better system performance when dealing with large-scale problems, and have a great potential to deal with real-time traffic management and control problems.