Project Description
Summary
This project aims to promote sustainable urban infrastructure and reduce reliance on private vehicles. By leveraging the power of geospatial data and Generative AI, the solution provides policy recommendations tailored to specific locations. The web application allows users to explore various datasets and receive AI-driven insights on public transport, electric vehicle infrastructure, and human mobility patterns, empowering better decision-making.
Problem Statement
In Australia's growing urban areas, developing sustainable infrastructure and reducing dependence on private vehicles is crucial. This project addresses the challenge through three main aspects:
Spatial Data Insights from Multiple Data Sources
Australia offers extensive geospatial data, ranging from local to state-level sources. However, integrating diverse datasets requires robust workflows to extract, transform, and unify data for meaningful analysis.
What AI tools help to solve the geospatial problems
While AI is commonly used for tasks like image recognition, the application of Generative AI for decision-making in geospatial contexts is emerging. Leveraging AI tools to solve geospatial challenges opens up new opportunities for innovation.
The spatial correlation between human mobility with public transport
As cities expand, increased car use worsens congestion and environmental impacts. Understanding how human mobility aligns with public transport systems is crucial for creating liveable cities. Spatial correlations between these factors can drive better planning and promote sustainable transport alternatives.
Solution Overview
Our proposed solution is a web application that integrates geospatial data and Generative AI to deliver policy recommendations for a given area. The system will provide tailored insights into public transport, EV charging stations, and human mobility patterns, helping decision-makers create more sustainable cities.
Web Application Architecture
The web app will feature an intuitive interface that allows users to select areas of interest by drawing a bounding box on a map. Users will input specific queries, and the app will provide AI-generated recommendations based on the available geospatial datasets. These recommendations will focus on improving transport infrastructure and sustainable city planning.
Geospatial Data and Datasets
Key datasets will be used to support policy recommendations, including:
- EV Charging Stations: Information on the availability and usage of EV charging points.
- Public Transport Usage: Data on route frequency, passenger capacity, and network coverage.
- Human Mobility: Data on traffic flow, commuting patterns, and movement within urban areas.
These datasets will be processed through a data workflow, ensuring the information is clean, relevant, and ready for analysis by AI models.
Interactive Map and Bounding Box
Users will interact with a dynamic map where they can:
- View geospatial data overlaid on the map.
- Draw a bounding box to select a specific area for analysis.
- Enter a query, such as “How can public transport be improved in this area?”
The AI will then analyse the selected area and provide recommendations based on the data within the bounding box.
AI-Driven Policy Recommendations
Powered by models like ChatGPT, the AI will interpret the user’s query and the geospatial data to offer policy suggestions. For example, the AI might recommend expanding public transport coverage in areas of high traffic congestion or suggest additional EV charging infrastructure where adoption is increasing.
These recommendations will be presented to users in a clear, text-based format that supports data-driven decision-making.
Technology Stack
Our solution uses a variety of tools to manage data, build the web app, and collaborate effectively:
1. Data Management
- Excel: For viewing datasets.
- Python & C#: For processing and analysing data.
- QGIS & geojson.io: For visualising and validating geospatial data.
2. Web App Development
- Nx & Angular: To build the front-end interface of the web app.
- Angular Material: The UI library for the web app.
- Leaflet: For creating interactive maps and displaying geospatial data.
3. Artificial Intelligence
- AI Model: Using a combination of the data available on the map and the user’s prompt, we can use AI models to generate policy recommendations to help with decision making
4. Workflow Diagrams
- Canva: For creating diagrams to visualise the app’s architecture and workflows.
5. Prototype
- Figma: For creating Prototype
6. Collaboration Tools
- Slack & Teams: For communication and team collaboration.
- Google Drive: For sharing files and storing project resources.
- Notion: For task management and documentation.
This tech stack ensures efficient data handling, user-friendly web app development, and smooth team collaboration.
Impact and Benefits
This solution offers significant benefits for sustainable urban planning by providing policymakers with data-driven, place-based recommendations. The use of Generative AI ensures that decisions are informed by the latest available data and tailored to specific locations, ultimately helping to:
- Promote public transport usage.
- Reduce private vehicle dependency.
- Support the transition to cleaner, greener urban infrastructure.
Challenges and Learnings
Throughout the project, integrating diverse data sources from various councils and state-level repositories posed a challenge. Developing a robust ETL workflow to clean and standardise this data was essential. Additionally, refining AI models to interpret geospatial data effectively required continuous learning and adjustments.
Future Improvements
To enhance the solution, future iterations could:
- Incorporate visual AI suggestions on the map, making the recommendations more intuitive.
- Allow users to customise how the map is symbolised based on specific criteria, enabling better visual analysis.
Resource
Social Media
Data Story
Data Workflow
Based on our experience, utilising the GeoJSON format for geospatial data significantly enhances webpage performance and provides a smoother user experience. Additionally, it simplifies the development process. As a result, we opted to convert all geospatial datasets into the GeoJSON format.
To begin, we discovered that some datasets from the digital atlas were available via APIs that directly provided data in GeoJSON format. We leveraged these APIs to retrieve the data efficiently.
The data from Data Vic was categorised into three main parts:
- Excel and CSV files: For these datasets, we used scripts written in C# and Python to clean and preprocess the data, preparing it for conversion.
- Shapefiles: We used QGIS to convert these shapefiles into the GeoJSON format.
- Existing GeoJSON files: Datasets already in GeoJSON format were directly integrated without further conversion.
Once all data was consolidated in the GeoJSON format, we performed validation and preliminary visualisation using geojson.io and QGIS to ensure accuracy and completeness.
Insights and Outcomes
We primarily collected the following data: Local Government Area boundaries, public transport, road infrastructure, information about cars and electric vehicles, bicycle infrastructure data, and population density. Supported by this data, and with the aid of visualisation tools, we are able to identify the intrinsic relationships and spatial correlations between residents, travel modes, urban transport, and road infrastructure across one or more regions. With the integration of AI, users can quickly and accurately obtain the desired information regarding population movement from vast datasets.
Conclusion
Wrap up by reiterating the value of data in driving smarter, more sustainable decisions for solar energy development, and how your project makes that possible in a user-friendly, interactive way.
