Vision-based Obstacle Detection and Motor Speed Control for Autonomous Driving Systems

Abstract

Autonomous driving systems rely on robust perception and control mechanisms to navigate safely in dynamic environments. This study presents a vision-based approach for obstacle detection and motor speed control using MobileNet SSD object detection model. The system utilizes a camera module to capture real-time video frames, which are processed to detect and classify obstacles. Based on the detected objects' position and distance, an adaptive motor speed control algorithm adjusts the vehicle's velocity to ensure collision avoidance and smooth navigation. The implementation is tested on a Raspberry Pi-based platform with an integrated motor control system, utilizing PWM signals for speed regulation. MobileNet SSD offers a lightweight, faster alternative for real time inference. Experimental results demonstrate the system’s effectiveness in detecting obstacles and dynamically adjusting speed in response to environmental conditions. This approaches enhance autonomous vehicle safety and efficiency, making it suitable for real-world applications in self-driving technologies.