The ASSA Project: An Intelligent Agent Simulation of Shop-Around Behavior

Policy exploration for downtown revitalization gets becoming important in also Japan. This article deals with a modeling and simulation project of downtown visitors' shop-around behavior using an intelligent agent approach, by mainly devising and implementing planning (dynamic scheduling/ rescheduling), learning and adaptation functions. In this article, first, we explain the architecture of ASSA model. Each agent makes and remakes his/her schedule to visit shops based on time constraints and shop preferences, chooses alternative visits when he/she fails in an errand, and makes impulse stops at shops and detour actions. ASSAver.3 was reported based on detailed observations and surveys taken at Asunal Kanayama shopping mall, and Osu shopping district, Nagoya. Then we show a framework of evaluation that included redundancy indicators for shop visits and walk lengths, and similarity analysis of shop-visit sequences. Valid-check results were shown in the simulation performances of the two cases of from these three aspects.