This project-based seminar provides master’s students with hands-on experience in the complete workflow from physical Lego system to calibrated Digital Twins. Working in small teams, students construct an agent-based system with Lego SPIKE Prime, design and execute data collection pipeline, and apply data-driven techniques to extract agent behaviors and model parameters. The extracted model is then used to construct and validate a Digital Twin, which is subsequently used for simulation experiments and policy exploration. The seminar bridges the gap between physical agent behavior and virtual simulation, equipping students with transferable skills in data-driven modeling, system identification, and simulation-based analysis.

Learning Objectives

By the end of this seminar, students will be able to:

• Understand the relationship between physical agent-based systems and their Digital Twin representations
• Design and implement data collection and preprocessing pipelines for agent-based systems
• Select and apply appropriate data-driven parameter extraction techniques (e.g., curve fitting, inverse modeling, machine learning)
• Build and calibrate a Digital Twin simulation from empirically extracted parameters
• Validate Digital Twin fidelity using quantitative behavioral metrics (e.g., RMSE, attack rate replication)
• Design and execute simulation experiments to explore policy scenarios and intervention strategies
• Evaluate model performance and articulate limitations with respect to real-world applicability
• Communicate technical findings effectively through structured presentations and written reports

Required Skills

• Object-oriented programming (Python, Java, or equivalent)
• Basic understanding of data analysis and preprocessing
• Familiarity with agent-based modeling concepts
• Basic knowledge of statistics (mean, variance, hypothesis testing)

Size:  8 Students (4 per group)

Deliverables: 1 Report per Team + 3 Presentations + Implementation files (Video, Code, Datasets, etc.)

Course Components: 2 Lectures + 3 Presentations
Lecture 1: Introduction to Project and Data-driven Agent-Based Modeling and Digital Twins
Lecture 2: Data Collection, Parameter Extraction, and Validation Methods for Digital Twin Construction