Author: Purba Daru Kusuma, Azhari Azhari, and Reza Pulungan (2016)
Problem and Challenge
Traditional markets in Indonesia are facing a decline due to the rise of modern retail environments. Existing crowd simulation models—often focused solely on movement—fail to capture the nuanced behaviors of buyers in traditional markets, such as spontaneous purchasing and interactions with traders that block corridor space. These limitations make it difficult to design effective crowd management strategies for revitalizing traditional markets.
A diagram showing a narrow market corridor with static trader booths and dynamic buyer agents interacting and obstructing flow (Fig. 1).
Goal of Experimentation
The goal is to develop a crowd simulation model tailored to traditional markets that incorporates both movement and decision-making behaviors of buyers. By simulating real-world conditions in markets like Gedongkuning and Ngasem, the model aims to help stakeholders understand spatial dynamics and optimize market layouts for better crowd flow and customer experience.
Methods
The simulation uses a hybrid approach combining cellular automata for pedestrian dynamics and social force models for interaction behavior. Each buyer is modeled as an intelligent agent with preferences, loyalty levels, and responsiveness to trader attractiveness. The environment is represented as a 2D grid, where agents navigate based on available space and decision-making logic.
System Architecture
The propose of system architecture is shown in Fig.2: The diagram shows the multi-agent simulation platform that models traditional market dynamics through intelligent, autonomous agents. With BDI capabilities, agents interact, collaborate, and replicate in response to their environment, generating realistic behaviors. The system’s flow—from perception to decision to market action—merges AI precision with human-like complexity. Its warm, minimal design evokes trust and community, making SimMarketAI both a powerful tool and a relatable reflection of real-world market life.
Figure 2. System Architecture
Result and Discussion
Simulations revealed that certain areas of the market are more frequently visited, aligning with real-world observations. The model successfully captured dynamic buyer behavior, including spontaneous changes in purchasing targets. It demonstrated that crowd density is influenced by market layout, trader attractiveness, and buyer arrival intervals. The results validate the model’s realism and its potential for aiding in traditional market revitalization.
Figure 3. Crowd Simulation by Intelligent Agents
Value Proposition
SmartCrowdSim is an innovative simulation platform that blends agent-based modeling with AI to optimize traditional market ecosystems. It enables planners, retailers, and policymakers to simulate buyer loyalty, trader appeal, and crowd movement—offering data-driven insights for layout design, vendor placement, and behavioral forecasting. Ideal for smart city planning, retail strategy, and academic research, SmartCrowdSim transforms informal market dynamics into actionable models, bridging tradition with technology for scalable, real-world impact.
