Dynamic Urban Sandbox - Interactive Policy Simulation

Freestyle Theme - Simulate policy impacts on Hong Kong's urban development through AI agent-based modeling, giving more insights in making an effective policy to enhance quality of life of public.

Comment

Inspiration

With the rapid advancement of technology, smart city is receiving increasing attention. We have discovered that government is now concerning the development of "Dynamic Policy". Therefore, it inspired us to create a platform which allows government or any other administrators test their policy rather than directly deploy it in reality to collect the data, helping them to reduce the time and cost of establishing a policy. Thus, we would like to focus on "Smart Government".
A Good Policy Affect the Environment, and the Environment Affect Citizen Lives

What it does

🏙️ City Simulation

  • AI Agent Citizens: 1000 simulated Hong Kong citizens with diverse backgrounds (age, occupation, income, district)
  • Dynamic City Model: Visual representation of Hong Kong districts with different zone types
  • Real-time Data Integration: Simulated data collection and correlation analysis

🤖 AI Agent System

  • Individual Decision Making: Each citizen AI agent evaluates policies based on personal characteristics
  • Interactive Behavior: Agents influence each other's decisions, creating realistic social dynamics
  • Adaptive Responses: Agents adjust their support based on age, income, occupation, and personal priorities

📊 Policy Analysis

  • 5/10/15/20-Year Simulation: Projects long-term impacts of proposed policies
  • Support Metrics: Shows percentage of citizen support vs opposition
  • Decision Criteria: Reveals the main factors AI agents consider when evaluating policies
  • Vulnerability Assessment: Identifies potential weak points and provides actionable recommendations

How we built it

We built the sandbox based on the following technical implementations:

AI Agent Modeling

  • Citizen Characteristics: Age (20-80), occupation, income (20k-120k HKD), district
  • Decision Factors: Personal priorities (housing, transport, environment, economy, healthcare, education)
  • Policy Evaluation: Multi-factor analysis considering demographics and policy content

Simulation Process

  1. Data Collection: Simulates gathering Hong Kong real-time data
  2. Model Generation: Creates city model with diverse AI agent population
  3. Policy Simulation: Runs 10-year impact analysis with agent interactions
  4. Results Analysis: Generates support metrics, criteria, and recommendations ## Challenges we ran into

Accomplishments that we're proud of

we have a clear mindset that what a whole project we are going to build. Before using Amazon Q to write program to build a demo, we have made a PPT proposal for ourselves to summarize what would be our motivation of creating such a project, what is the sandbox's operational processes, how it works, what benefits it can bring to society. Actually, this is also the challenges we faced before we start to build the demo project because if we don't know what we want to make, it would be difficult to create a demo.

By finishing such a proposal, we fully understand our project and know what functions we want Amazon Q to build.

What we learned

We have applied such an opportunity to adopt what we have learnt about IT knowledge into making a sandbox. While this sandbox can bring benefits to the society, we found that our ideas may have a chance to enrich our society. This motivates us to use our creativity more in the future, thinking more good projects to improve Hong Kong.

What's next for Dynamic Urban Sandbox - Interactive Policy Simulation

We have a future plan if the time and budgets are available for us to further develop this sandbox:

Future Enhancements

  • Integration with real Hong Kong government data APIs
  • More sophisticated agent interaction models
  • Advanced visualization with 3D city representation
  • Deploy a real LLM API, using machine learning for improved policy impact prediction (e.g. RAG)
  • Flexible selection of Hong Kong different cities to test a policy
  • Use more engines to support the sandbox development (e.g. Unreal Engine)

Built With

Share this project:

Updates