Development of auto controlled traffic light using machine learning

M. Rummaja, Mohd Iskandar Dzulkarnain (2021) Development of auto controlled traffic light using machine learning. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

	Text (24 Pages) Development of auto controlled traffic light using machine learning.pdf - Submitted Version Download (1MB)
	Text (Full text) Development of auto controlled traffic light using machine learning.pdf - Submitted Version Restricted to Registered users only Download (6MB)

Abstract

As the city's population and automotive traffic grow, traffic congestion at intersection is becoming a significant issue in many large cities. The inefficiency of the existing traffic light system's approaches and algorithms is one of the causes of the traffic issue. Therefore, adopting new systems and technologies capable of improving the current traffic light control system is urgently needed to prevent the situation from worsening. This project proposes an automatically controlled traffic light based on the machine learning technique. The concept used to develop an auto-controlled traffic light that be control the traffic congestion. The system was developed on two lanes of a four-legged intersection with a turning signal. By using a machine learning technique, a reinforcement learning algorithm, the system aims to providing the optimal traffic light phase and adjusting the timing of the green phase, depending on vehicle demand. Furthermore, vehicle demand was measured by an induction loop sensor implemented on the road along the intersection. Based on the simulation result, the proposed system be able to altering the green light duration and providing a suitable traffic light phase sequence based on the road congestion compared to static traffic light, which has a fixed green light system and traffic light phases.

Actions (login required)

View Item