Smart Lights Detection System for Automotive Safety

The Challenge

PERODUA, Malaysia's leading automotive manufacturer producing over 350,000 vehicles annually, faced critical quality control challenges in their lighting system verification process:

Manual Inspection Bottlenecks: Quality control teams were manually inspecting every vehicle's lighting system, creating production delays and inconsistent results.
Human Error Risk: Visual fatigue from repetitive inspections led to a 2-3% error rate in detecting faulty installations or non-functioning lights.
Production Line Speed: Manual checks limited line speed to 60 vehicles per hour, below the target of 100 vehicles per hour.
Compliance Pressure: Stringent automotive safety regulations required 100% verification of all lighting systems with documented proof.
Cost Inefficiencies: Large QC teams and rework costs were impacting operational margins.

"Safety is our top priority at PERODUA. We needed a solution that could guarantee 100% accuracy in detecting lighting defects while maintaining our production efficiency. Manual inspection was no longer sustainable."
- Dato' Zainal Abidin Ahmad, President & CEO of PERODUA

Our Solution

Fylix AI developed a comprehensive computer vision system specifically designed for automotive lighting quality control:

1. Multi-Camera Vision System

We deployed a sophisticated array of high-resolution cameras and sensors:

8 strategically positioned 4K cameras for complete vehicle coverage
Specialized low-light sensors for detecting minimal luminosity variations
Infrared cameras for heat signature validation of functioning lights
360-degree coverage ensuring no blind spots

2. AI-Powered Detection Algorithms

Our custom deep learning models were trained on:

Over 500,000 images of correct and faulty lighting installations
Pattern recognition for 47 different PERODUA models and variants
Real-time anomaly detection for alignment, brightness, and color temperature
Predictive maintenance indicators based on lighting patterns

3. Integrated Quality Management System

The solution seamlessly integrated with PERODUA's existing infrastructure:

MES Integration: Direct connection to Manufacturing Execution System for real-time updates
Automated Documentation: Every inspection automatically logged with photographic evidence
Alert System: Instant notifications to line supervisors for detected issues
Analytics Dashboard: Real-time quality metrics and trend analysis

4. Edge Computing Architecture

To ensure minimal latency and maximum reliability:

On-premise edge servers for real-time processing
Redundant systems ensuring 99.99% uptime
Sub-100ms detection time per vehicle
Offline capability for uninterrupted operations

Implementation Process

The implementation was carefully planned to minimize production disruption:

Phase 1: System Design and Testing (Weeks 1-6)

Conducted detailed analysis of all PERODUA models and lighting configurations
Designed custom mounting solutions for camera arrays
Built test environment replicating production line conditions
Developed initial AI models using historical quality data

Phase 2: Pilot Installation (Weeks 7-10)

Installed system on one production line during scheduled maintenance
Parallel running with manual inspection for validation
Fine-tuned algorithms based on real production data
Trained quality control staff on new system operation

Phase 3: Full Deployment (Weeks 11-14)

Rolled out to all 5 production lines over successive weekends
Implemented 24/7 monitoring and support protocols
Established continuous learning pipeline for model improvement
Created comprehensive documentation and training materials

Phase 4: Optimization and Expansion (Weeks 15-16)

Fine-tuned detection parameters based on production feedback
Expanded system capabilities to detect other visual defects
Integrated predictive analytics for maintenance planning
Achieved full production speed targets

Key Milestones

Week 1 Project Kickoff & Requirements Analysis

Week 6 AI Models Trained & Validated

Week 10 Pilot Success - 99.8% Accuracy Achieved

Week 14 Full Production Deployment

Week 16 System Optimization Complete

Results & Impact

The Smart Lights Detection System delivered exceptional results across all key metrics:

Quality & Accuracy Improvements

99.8%

Detection Accuracy

From 97% manual inspection accuracy

85%

Defect Reduction

Near-zero defects reaching customers

100%

Inspection Coverage

Every vehicle thoroughly checked

Operational Efficiency Gains

Production Speed: Increased from 60 to 110 vehicles per hour, exceeding targets by 10%
Labor Optimization: Reduced QC team size by 50%, reassigning staff to higher-value activities
Processing Time: Inspection time reduced from 3 minutes to 20 seconds per vehicle
Rework Costs: 75% reduction in post-production fixes, saving RM 2.5 million annually

Business Impact

💰 Cost Savings

RM 4.2 million annual savings from improved efficiency and reduced defects

🏆 Quality Awards

Received Malaysia Automotive Excellence Award for Quality Innovation

📈 Customer Satisfaction

Quality-related complaints dropped by 92%, improving brand reputation

Strategic Advantages

Beyond immediate metrics, PERODUA gained significant competitive advantages:

Industry-leading quality assurance capabilities setting new benchmarks
Data-driven insights enabling proactive quality improvements
Enhanced regulatory compliance with automated documentation
Foundation for expanding AI across other production processes

"This AI system has transformed our quality control processes. We're not just detecting defects faster – we're preventing them. The data insights have helped us improve our entire manufacturing process. It's a game-changer for PERODUA."
- Mohd Azizi Mohd Noor, Head of Manufacturing