UTM FYP — PSM1 / PSM2 Proposal

ORBIT

On-Demand · Route-Based · Intelligent Transit

A system that gives UTM's buses the ability to think — and gives students the information they actually need.

Large
Campus
UTM Skudai
19Bus Services
3Layers of Solution
0New Routes Needed
Scroll

Students wait. Buses run.
Nobody knows why it's broken.

UTM's campus spans approximately 1,222 hectares. You can't walk from your college to your faculty. You depend entirely on the bus — but the bus has no idea you're waiting.

Central Problem
UTM's campus shuttle services operate on fixed schedules with no mechanism to detect or respond to real-time passenger demand — leaving students stranded at stops during peak periods with no information on bus location, arrival time, or seat availability.
⏱️

Buses Can't See Demand

Every bus follows a fixed clock — 3 students or 40 students at the stop, the bus behaves identically. There is no feedback between real demand and the schedule.

Extended service gaps during pre-lecture peaks
📵

Students Are Left Guessing

No app, no screen, no sign at the stop tells you when the next bus comes or how full it is. You stand, you wait, and you hope. The myUTM app has no bus information at all.

Zero bus tracking in myUTM today
🪑

Fleet Management is Blind

There is no live view of queues building, buses running early or late, or corridors under pressure. Problems are discovered after they happen — not while they're happening.

No real-time fleet dashboard exists
Why We Know This Is Real at UTM
📄

Funded UTM Research Initiative

A RM188,127 UTM-funded research project (2026–2028) formally documents the same problem: overcrowding, unpredictable bus frequency, and students at KDOJ and KDSE specifically burdened by long waits with no notifications.

🗓️

Official UTM Bus Schedule

The published UTM bus schedule confirms all 19 services operate on fixed timetables with no demand-sensing mechanism of any kind — arrival times are explicitly stated as not guaranteed.

🎓

Student Consultation

Preliminary consultation with UTM students who regularly use campus shuttle services confirmed recurring experiences of extended waits during pre-lecture periods and no access to real-time bus information.

Three layers.
One connected system.

Each layer solves a different piece of the puzzle. Together, they form a smart, demand-aware campus transit system — built on top of the routes that already exist.

1
For Students

The Student App

Students open a simple page on their phone at the bus stop and tap where they want to go. That's it. In return, they immediately see when the next bus arrives, how full it is, and where it last was. No download. No login. Works on any smartphone.

🚌 Real-time ETA 🟢 Capacity View 📍 No App Install 🔒 Fully Anonymous
2
The Intelligence

Smart Coordination Behind the Scenes

Every bus and every bus stop is an autonomous agent — a tiny decision-maker that operates independently. When a stop detects a big queue building, it sends a signal. The right bus evaluates: leave a little earlier? Hold and wait? Flag the problem for management? All automatically, in real time, with every decision explained and logged.

🤖 Bus Agents 🤖 Stop Agents 📡 Live Messaging ⚡ Millisecond Decisions
3
For Fleet Management

The Fleet Dashboard

Fleet managers get a live view of the entire campus network — every bus, every stop, every queue, right now. They can see exactly what decision the system made at any moment and why. And they can override any bus manually at any time. For the first time, UTM Fleet has complete operational visibility.

🗺️ Live Campus Map 📋 Full Decision Log 🎛️ Manual Override 📊 Fleet Metrics

GPS tracking makes buses visible.
ORBIT makes them intelligent.

A funded UTM research initiative is already deploying GPS tracking infrastructure across all campus corridors. ORBIT is the coordination intelligence layer above it — not competing with it, complementing it.

UTM IoT-GPS Initiative (Active, 2026–2028)

Bus Visibility

  • GPS trackers on every bus
  • Real-time bus location for passengers
  • ML-based ETA prediction
  • Display screens at bus stops
  • No student demand sensing
  • No autonomous dispatch decisions
  • Buses still run fixed schedules
ORBIT — This Project

Bus Intelligence

  • Students signal demand at stops
  • Agents detect queue build-up in real time
  • Buses adjust timing autonomously
  • Fleet manager sees every decision and why
  • Designed to consume GPS feed when live
  • No new routes, no new hardware needed
Why It Matters

What actually changes
when buses become smart.

Shorter Waits at Peak Hour

Buses respond to real demand, not just the clock. Long gaps during the rush before lectures are detected and corrected automatically.

📲

Students Finally Know

ETA, seat availability, last known bus position — delivered on any phone browser, right at the stop, with no friction.

🧠

Coordination Without a Boss

Agents work together without a central controller. No single point of failure. If one part breaks, the rest keeps running on schedule.

🗺️

Full Fleet Visibility

For the first time, management can see the whole network live — and get a complete explanation for every decision the system made.

🛣️

Zero Disruption to Operations

No routes change. No buses are moved. Only departure timing is gently adjusted, within strict safety limits, when justified.

📈

Gets Smarter Over Time

Every student tap builds a richer picture of real demand patterns on campus. The more students use it, the more accurate the system becomes.

📡

IoT-Native from Day One

Student phones, bus GPS trackers, and future smart sensors at stops all speak the same protocol — MQTT, the backbone of modern IoT. The system is ready for physical hardware without redesign.

🤖

AI That Actually Makes Sense

A lightweight AI model watches incoming demand signals and flags anything that looks abnormal — protecting the system from bad data before it can trigger a wrong decision.

🌍

Built for More Than UTM

Every UTM-specific detail lives in a config file. Any university, hospital shuttle, or corporate bus operator can deploy the same system by swapping config — no code changes needed.

Built for everyone on campus.

🎓
The Student

Less guessing,
more getting there.

  • Know the next bus ETA before even leaving your room
  • See how full the bus is before walking to the stop
  • Never miss a lecture because of an unexpected gap
  • Works on any phone — no app download, no login
🚌
UTM Fleet Management

See everything.
Act fast.

  • Live view of all 19 services on one screen
  • Every system decision logged with full reasoning
  • Override any bus manually at any time
  • The first real-time operational tool UTM Fleet has ever had
🏫
UTM as an Institution

Modern campus,
smarter service.

  • Real student demand data — not guesses — for fleet planning
  • Evidence base for future investment decisions
  • No new routes, no new buses required to deploy
  • Expandable to all 19 services after initial validation
The Plan

Design this semester.
Build it next year.

PSM1 is the full blueprint. PSM2 is the construction — layer by layer, in order of student impact. The system has standalone value at every stage.

PSM1 — Now

The Blueprint

Full system design: architecture, agent behaviour rules, app screens, database structure, and every design decision documented with honest reasoning.

Chapters 1 – 5
PSM2 — Phase 1

Student App First

Build the student-facing layer first — it has standalone value and produces the real demand data the agents need. Working on a real phone by Week 6.

Demo 1 — 40%
PSM2 — Phase 2 & 3

Agents + Dashboard

Add the smart coordination layer and fleet dashboard. Full three-layer system running live by Week 15, verified against defined test scenarios.

Final Demo — 100%
PSM2 Scope — What Keeps This Manageable
3

Bus corridors modelled in simulation — not all 19. Bus B, E, and F. Representative, not exhaustive.

0

Physical hardware on real buses. The entire PSM2 evaluation runs on software-in-the-loop simulation with a virtual data injector.

3

Layers built in priority order. Even if the dashboard is incomplete, the app and MAS backend constitute a working, demonstrable system.

1yr

Head start. PSM1 approval is now. PSM2 build semester starts next academic year — preparation begins immediately.

Proven tools.
IoT & AI included.

All core components are free and run locally. IoT hardware marked with * is Future Work — the architecture already supports it through a pluggable sensor interface.

Student App
React PWA Geolocation API WebSocket FastAPI
Agent Backend
Python MQTT Mosquitto NetworkX Dijkstra
Fleet Dashboard
React Leaflet.js WebSocket
IoT Layer
MQTT Protocol GPS Sensing mmWave Radar* APC Sensors*
AI Component
Isolation Forest Anomaly Detection scikit-learn
Data & Testing
SQLite pytest 16 Test Cases Black Box Testing

Built for UTM.
Designed for anywhere.

Every UTM-specific detail — stop names, routes, schedules, capacity — lives in a config file. Any transit operator swaps their own config and the system works. No code changes. No redesign.

🏫 Any University 🏥 Hospital Shuttles 🏭 Corporate Buses 🏙️ Any Fixed Route