IndySafetyAssist_bot

• 

  How bot Reacts to /my-reports command

• 

  Bot initial start

🔥 Inspiration The increasing number of public safety issues and lack of accessible communication channels inspired us to build a solution that empowers citizens to instantly report hazards and receive timely alerts. We envisioned a lightweight, user-friendly assistant that bridges the gap between authorities and the public—available right from a familiar platform: Telegram.

🚀 What it does . IndySafetyAssist is a smart Telegram bot that: . Allows users to report local hazards by sharing their location, description, and urgency and all necassary deatils using /report_hazard command. . Automatically classifies the type of hazard using a trained ML model. . Notifies users with real-time safety alerts based on their location,which they had set using /set_location command. . Enables users to track the status of their reports and view detailed updates using /my_reports command. . Provides a two-way communication channel between users and authorities/admins. . Admin can update the report status and can send targetted alerts using our secure Admin Api endpoint. . User can ask questions about emergency situation and our bot can answer for it using our knowledge base with simple NLU model.

🛠️ How we built it

. Frontend: Telegram Bot using python-telegram-bot. . Backend: Flask API hosted on Google Colab for ML classification and MongoDB Atlas for storing reports, alerts, and user data. . ML Model: A trained classifier to predict the type of hazard based on description.

Features Integrated:

. Inline buttons for interaction (View Report, Back, etc.). . MongoDB for persistent storage. . Location-based filtering for alerts. . Admin endpoints for updating statuses and sending alerts.

🧗 Challenges we ran into

. Handling dynamic inline keyboard interactions reliably. . Managing the flow of user inputs and maintaining state across multiple steps in Telegram. . Ensuring accurate hazard classification using a limited dataset. . Implementing real-time, location-aware alert distribution. . Debugging Telegram callbacks, especially inline button updates and message edits. . Integrating ML model to the python bot code.

🏆 Accomplishments that we're proud of

. Created a fully functional bot that simulates a real-world public safety assistant. . Implemented ML-based hazard classification and integrated it into the reporting flow. . Achieved smooth multi-step user input collection with error handling. . Built a working prototype that supports both citizen and admin use cases.

📚 What we learned

. Advanced usage of the python-telegram-bot library for handling conversations and inline callbacks. . Integration of ML models into real-time applications using Flask APIs. . MongoDB document structuring for scalable user and report data. . Designing intuitive user flows for bots on mobile chat platforms.

🚧 What's next for IndySafetyAssist_bot

. Add photo-based classification using computer vision models. . Implement voice-enabled reporting for accessibility. . Create an admin dashboard to visualize reports and alerts geographically. . Add community voting to validate reports and reduce spam. . Integrate emergency contact routing and notifications to nearby services.

Built With

• colab
• dnspython
• flask
• machine-learning
• mongodb
• ngrok
• nlu
• numpy
• pandas
• postman
• python
• python-telegram-bot
• scikit-learn
• sentence-transformer
• telegram-api
• webhook