Prototype of Duckietown Car Robot Control System on Traffic Light

Abstract

This study aims to design and develop a prototype control system for the autonomous Duckiebot robot that can detect and respond to traffic lights in the Duckietown arena. Duckiebot uses Raspberry Pi 3 as a controller and a fisheye camera as a sensor to detect objects around it, including traffic lights. This system implements lane-following mode, where Duckiebot follows a predetermined path until it detects a stop line. After stopping right in front of the stop line, Duckiebot uses the camera to read the color of the traffic light and take appropriate action. If a red light color is detected, Duckiebot will stop, while if a green light color is detected, Duckiebot will continue the journey. Testing was conducted to evaluate Duckiebot's ability to detect traffic lights under various conditions, including detection with different lighting and when the traffic light is only partially visible. The test results showed that Duckiebot had a success rate of 70% in detecting traffic lights with appropriate responses, although the camera's sensitivity to lighting is an important factor affecting detection accuracy. This research is expected to contribute to the development of autonomous robot control systems for smart transportation applications and future robotics research.