Safety-first mobility intelligence

Safer Journeys with SafeDriver

SafeDriver helps passengers, drivers, and operators make every journey more predictable and secure with real-time driver telemetry, live route tracking, AI predictions, and Google Maps integration.

Discover Features View Implementation

24/7

Live telemetry

Route confidence

Real-time

Safety alerts

Live route visibility

Location updates and journey context for safer public transport.

AI risk detection

Fatigue, distraction, and behavior cues analyzed in real time.

Instant safety alerts

Fast notifications help operators respond before risk escalates.

Our Domain

Discover how SafeDriver is improving public transport safety through AI-powered real-time driver monitoring and intelligent accident prevention.

Research Background

Road traffic accidents remain a major safety challenge in Sri Lanka, particularly in public transportation where driver fatigue, distraction, and unsafe driving behaviors contribute to many preventable incidents. Existing monitoring approaches rely heavily on manual supervision and post-accident investigations, creating a need for an intelligent system capable of continuously monitoring drivers and preventing accidents before they occur.

Research Problem

Public transport authorities currently lack a reliable real-time solution to detect driver drowsiness, mobile phone usage, smoking, and other dangerous behaviors while buses are in operation. Without immediate alerts and centralized monitoring, unsafe driving conditions often go unnoticed until accidents occur, putting both passengers and drivers at significant risk.

Methodology

SafeDriver utilizes an AI-powered edge-cloud architecture built with Raspberry Pi 4, OpenCV, MediaPipe, TensorFlow Lite, and YOLO to detect driver fatigue and distractions in real time. The system integrates GPS tracking, Firebase cloud services, a Next.js authority dashboard, and Flutter mobile applications to provide intelligent alerts, fleet monitoring, passenger feedback, and comprehensive transport safety management.

Literature Survey

The literature survey indicates that although numerous AI-based Driver Monitoring Systems (DMS) have been developed worldwide, many existing solutions primarily target luxury or commercial vehicles and rely on expensive proprietary hardware or cloud-based processing. Most systems focus only on detecting driver drowsiness while providing limited support for distraction detection, real-time intervention, passenger engagement, and transport authority monitoring.

Recent studies demonstrate that computer vision techniques such as MediaPipe, OpenCV, YOLO, and TensorFlow Lite can accurately detect fatigue indicators including prolonged eye closure, yawning, head pose, smoking, and mobile phone usage. However, many solutions require high-performance hardware or continuous internet connectivity, making them less suitable for large-scale deployment in Sri Lanka's public transportation sector.

Modern research also emphasizes edge AI, embedded computing, GPS integration, and cloud analytics for improving road safety. Nevertheless, there remains a lack of affordable, locally adaptable systems that integrate embedded AI, intelligent alerting, authority dashboards, driver mobile applications, passenger applications, and real-time fleet monitoring into a single platform.

Research Gap

Despite significant advances in intelligent transportation systems, several important research gaps remain in existing driver monitoring solutions.

Many existing systems concentrate primarily on drowsiness detection while overlooking unsafe driving behaviors such as mobile phone usage, smoking, head-turn distraction, and prolonged inattentiveness.

Most commercial driver monitoring solutions rely on expensive proprietary hardware or cloud-based processing, increasing deployment costs and reducing performance in environments with limited network connectivity.

Current solutions rarely provide a complete ecosystem connecting drivers, passengers, and transport authorities. Driver mobile assistance, passenger feedback, GPS hazard mapping, centralized fleet monitoring, and compliance reporting are often unavailable within a unified platform.

There is a lack of affordable AI-powered driver monitoring systems specifically designed for Sri Lanka's public transport environment, supporting right-hand-drive vehicles, multilingual interfaces, real-time edge processing, and intelligent multi-level alert mechanisms.

Research Objectives

AI-Based Driver Monitoring

Develop an embedded AI system capable of continuously monitoring driver behavior and detecting drowsiness using Eye Aspect Ratio (EAR), blink frequency, yawning detection, and head pose estimation.

Intelligent Distraction Detection

Implement computer vision and deep learning models using YOLO, OpenCV, and TensorFlow Lite to identify unsafe behaviors including mobile phone usage, smoking, head-turn distraction, and driver inattentiveness in real time.

Multi-Level Alert System

Develop an intelligent alert mechanism that delivers graduated responses through buzzer alarms, voice guidance, vibration alerts, driver mobile notifications, cloud alerts, and authority notifications.

GPS-Based Hazard Monitoring

Integrate GPS technology to identify accident-prone locations, monitor bus routes, support hazard mapping, and improve transport safety through location-aware analytics.

Driver Mobile Application

Develop a Flutter-based driver application that provides real-time safety alerts, trip information, driving performance insights, emergency notifications, and communication with the cloud platform.

Passenger Mobile Application

Develop a cross-platform Flutter application enabling passengers to monitor bus safety status, submit driver and bus feedback, access emergency SOS services, scan QR codes, and earn reward points for active participation.

Authority Dashboard

Build a centralized Next.js web dashboard that enables transport authorities to monitor drivers and buses in real time, manage alerts, visualize fleet activity, review passenger feedback, and generate compliance reports and analytics.

Cloud-Based Data Management

Implement Firebase Authentication, Firestore, Cloud Functions, Cloud Messaging, and secure cloud synchronization to support real-time communication, notifications, analytics, and centralized data management.

Platform capabilities

Key Features

Discover the AI-powered safety technologies that make SafeDriver an intelligent driver monitoring and accident prevention platform.

Driver Safety

AI Driver Monitoring

Continuously monitors driver alertness using AI-powered computer vision to detect fatigue, prolonged eye closure, yawning, and abnormal head movements in real time.

SafeDriver module01

Vision AI

Drowsiness Detection

Uses Eye Aspect Ratio (EAR), blink frequency, PERCLOS analysis, and facial landmark tracking to identify early signs of driver fatigue before accidents occur.

SafeDriver module02

Computer Vision

Distraction Detection

Detects unsafe driver behaviors including mobile phone usage, head-turn distraction, and inattentiveness using YOLO and MediaPipe running directly on the embedded device.

SafeDriver module03

AI Detection

Smoking Detection

Recognizes smoking behavior inside the vehicle cabin using deep learning models, helping improve driver discipline and passenger safety.

SafeDriver module04

Smart Alerts

Multi-Level Alert System

Provides intelligent safety interventions through voice guidance, buzzer alarms, vibration alerts, driver mobile notifications, cloud notifications, and authority escalation based on event severity.

SafeDriver module05

Edge Computing

Embedded Edge AI

Runs AI inference locally on Raspberry Pi 4 using TensorFlow Lite and OpenCV, ensuring low-latency detection even without continuous internet connectivity.

SafeDriver module06

Driver App

Driver Mobile Application

Provides drivers with live safety notifications, trip information, vibration alerts, driving history, emergency assistance, and performance summaries.

SafeDriver module07

Passenger App

Passenger Mobile Application

Allows passengers to monitor bus safety status, submit driver and bus feedback, scan QR codes, earn reward points, and access emergency SOS services.

SafeDriver module08

Fleet Monitoring

Authority Dashboard

A centralized Next.js dashboard enabling transport authorities to monitor buses, drivers, live alerts, routes, passenger feedback, and fleet performance in real time.

SafeDriver module09

Navigation

GPS Hazard Mapping

Tracks vehicle location, identifies accident-prone areas, visualizes hazard zones, and assists authorities in improving transport safety through location intelligence.

SafeDriver module10

Cloud Platform

Cloud Synchronization

Leverages Firebase Authentication, Firestore, Cloud Functions, and Cloud Messaging to securely synchronize alerts, driver data, analytics, and notifications across all connected devices.

SafeDriver module11

Reporting

Analytics & Compliance

Generates comprehensive reports on driver behavior, alert history, passenger feedback, safety trends, and regulatory compliance to support transport authorities in decision-making.

SafeDriver module12

Implementation Showcase

Visual implementation highlights from the SafeDriver embedded monitoring system, authority dashboard, passenger application, and driver application.

Embedded AI Unit

Monitoring System Implementation

Raspberry Pi 4-based monitoring system running real-time driver behavior detection with camera input, local alert devices, GPS data, and Firebase cloud transmission.