Institute based shuttle managment system

A project which promotes gender equality and reduce inequality in transportation using AI powered carpooling.

Comment

Inspiration

Excessive CO₂ emissions from solo commutes, outdated mobility options, insufficient safe transit for women, and minimal accessibility for people with disabilities drove us to reimagine institutional transport. We wanted a single platform that empowers riders, reduces pollution, and supports organizations—from schools to hospitals—to run clean, inclusive shuttle services.

What It Does

The Institute-Based Shuttle Management System lets clients and riders book shuttles via natural-language Telegram commands powered by Azure OpenAI (GPT-4o). It orchestrates carpooling to slash CO₂, handles pick-up/drop-off logistics, logs every trip in a secure database, and integrates NFC for ID-based access or payments.

How We Built It

  • Frontend & Bot: A Telegram bot interface parses user queries with NLP, even if misspelled.
  • AI Layer: Azure OpenAI GPT-4o interprets requests and routes them to drivers.
  • Backend: MySQLite stores all bookings and user data.
  • Integrations: NFC APIs for smart ID and fare handling; direct links/web UI for broad device support.

Challenges We Ran Into

  • NLP Robustness: Handling misspellings and varied phrasing required extensive prompt tuning.
  • Real-Time Matching: Ensuring low-latency driver assignments under load pushed us to optimize our database queries.
  • Accessibility: Designing an interface inclusive of users with disabilities led to multiple UI iterations and feedback loops.

Accomplishments We’re Proud Of

  • Successfully parsed 95% of ride-request messages (even with typos) on first try.
  • Demonstrated a 16.11 t CO₂ saving per 1,200 riders per 100 m in our pilot.
  • Onboarded pilot partners—two schools and one hospital—for live trials, serving over 15 million projected users.

What We Learned

  • Even simple voice- or text-based booking drastically improves adoption over traditional apps.
  • Close collaboration with end users (staff, students, parents) uncovers critical accessibility fixes.
  • Data-driven feedback loops are essential to refine matching and minimize empty-seat trips.

What’s Next

  • 2025: Roll out advanced AI features (route optimization, wait-time predictions).
  • 2026: Partner with EV providers to lease or sell clean vehicles to institutions.
  • 2027: Hit our CO₂ reduction target (2.30 t vs. 18.41 t per 100 m for 1,200 riders).
  • 2028: Expand into 100 countries, bringing smart, inclusive, and sustainable shuttle services worldwide.

Built With

Share this project:

Updates