Rekayasa Sistem Lampu Lalu Lintas berdasarkan Kepadatan Dua Persimpangan menggunakan Naive Bayes dan Komunikasi Soket

Abdul Muiz Anggit Budiyantoyo (Fakultas Ilmu Komputer, Universitas Brawijaya)
Eko Setiawan (Fakultas Ilmu Komputer, Universitas Brawijaya)
Mochammad Hannats Hanafi Ichsan (Fakultas Ilmu Komputer, Universitas Brawijaya)

Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer
Vol 5 No 11 (2021): November 2021

Congestion is a serious problem in this world. In Indonesia, congestion is a problem that always occurs every year. Congestion is caused by several factors, one of the causes of congestion is the ineffective and less optimal traffic light system that exists today. The current traffic light system often causes vehicles to accumulate in one of the sections that have a high level of vehicle density because the traffic lights on that section do not get the duration of the green light in accordance with the density of vehicles on that section. In addition to being caused by a high level of vehicle density, congestion in a segment is also caused by vehicles stopping on an active segment (green light) because they are waiting for a connecting link to a nearby intersection that is experiencing congestion so that vehicles cannot pass through the connecting segment. The system built by the author aims to optimize the current traffic light system to be more effective by providing a green light duration by considering the state of the density of vehicles on the section and also the density of vehicles on the section that connects the intersection with the surrounding intersection. This system uses a Raspberry Pi 4, TL LED light, and a Sumo Simulator running on a laptop. Roads and intersections are built and simulated using the Sumo Simulator. Naive Bayes is used to classify the Induction Loop Sensor data contained in the Sumo Simulator and predict the density level classes that exist on a road segment. The Raspberry Pi 4 will receive class prediction data that has been processed from a laptop using Socket Programming and will adjust the TL LED Light according to the data received. Tests are carried out on each segment that has been optimized using the system with an average accuracy of 91.66%. testing was also carried out using 3 different Route Trips with an average simulation result of 407 steps faster than the usual traffic light system. This system has an average execution time of 0.6057408571 seconds.