Design and development of an AI-based real-time road sign detection system for automatic speed control

The rapid growth in vehicular traffic has heightened the need for enhanced road safety and effective speed regulation. Speeding, particularly in sensitive areas like school zones and hospital vicinities, remains a major cause of accidents due to human error, low visibility, or disregard for signs. This paper presents the design and validation of an AI-assisted proof-of-concept system for real-time road sign detection and automatic speed control. The system integrates a Raspberry Pi 4B for image processing and an ESP32 microcontroller for speed regulation through Pulse Width Modulation (PWM). A high-resolution camera captures live video of the road ahead, where a Region-based Convolutional Neural Network (R-CNN) identifies potential traffic signs, and a Convolutional Neural Network (CNN) classifies them into specific categories such as school zone, hospital zone, or railway crossing. Upon detection, the system compares the vehicle’s current speed with the corresponding zone limit and automatically adjusts motor speed if overspeeding is detected, while a 16 × 2 LCD provides real-time feedback to the driver. Experimental testing under varied lighting and environmental conditions demonstrated promising detection accuracy, low-latency response, and effective speed regulation, confirming the feasibility of deploying deep learning models on low-cost embedded hardware. Rather than introducing new algorithms, this study focuses on the practical integration and performance evaluation of established CNN and R-CNN architectures within a resource-constrained environment. Future work will explore lightweight models and expanded datasets to further improve real-time performance and scalability in complex traffic scenarios.