Publication: Yapay zeka tabanlı araç hız tespiti
Abstract
Son yıllarda, dünya genelinde trafiğe kayıtlı araç sayısında hızlı bir artış gözlemlenmektedir. Bu artış, gelişmekte olan araç teknolojileriyle birleştiğinde, trafik yoğunluğu ve kaza riski gibi sorunların da ciddi biçimde arttığı belirlenmiştir. Kaza riskinin azaltılması amacıyla birçok ülkede çeşitli çalışmalar yapılmıştır. 1 Ocak 2004 tarihinde, karayolları ulaşımı ve bağlantıları üzerine araştırmalar yapmak amacıyla Ulaştırma Bakanları Avrupa Konferansı (ECMT) kurulmuştur. Dünya Sağlık Örgütü (DSÖ) ve ECMT'nin 2006 yılında birlikte hazırladığı Hız Yönetimi raporu, bahsi geçen alandaki önemli çalışmalardan biridir. Avrupa'daki birçok ülkede yapılan hız ve kaza önleyici çalışmaların ortak sonuçlarından biri, denetleme sistemlerindeki teknolojilerin kullanım oranının arttırılması gerekliliğidir. Elektronik sistemlerle karayolu denetiminin yapılması, saha elemanı ve araç kullanımını azaltarak daha pratik ve ekonomik bir yöntem sunacaktır. Bu sebep dolayısıyla, pek çok ülkede çeşitli elektronik denetim sistemlerinin kullanımı yaygın bir şekilde benimsenmektedir. Bahsedilen sistemler aracılığıyla hız sınırı ihlali, şerit ihlali, emniyet kemeri kullanımı gibi kontroller gerçekleştirilebilmektedir. Ülkemizde, elektronik denetim sistemlerindeki hız limitleri, kullanılan karayoluna ve araç tipine göre farklılık göstermektedir. Bu sistemler, farklı karayolu tipleri ve araç kategorileri için özel olarak belirlenmiş hız sınırlarını içermektedir. Araçların belirli bir mesafeden geçiş zamanlarıyla yapılan hız tahmini işlemi, Kalman filtresi kullanılarak test edilmiştir. Kontrol sürecinde, her araç benzersiz bir kimlik değeriyle tanımlanmış ve hız sınırı aşımı durumunda benzersiz kimlik numarası, ortalama hızı ve anlık yakalanan ekran görüntüleri kaydedilmiştir. Tespit edilen araçların hızlarını doğrulamak için Kalman filtresi ve manuel test aracı kullanılarak testler yapılmıştır. Yapılan testlerden alınan bir araç hız tespiti örneğinde; 70 km/ h sabit hızla yapılan bir geçişte 70.34 km/ h hız değeri gösterirken, Kalman filtresiyle hesaplanan hızın 70,27 km/ h, Kokusuz Kalman filtresiyle hesaplanan hızın 69,99 km/ h, Çok Değişkenli Kalman Filtresi ile hesaplanan hızın 70,02 olduğu tespit edilmiştir. Test aracıyla yapılan sabit hızlı geçişler sonucunda araç hızının tespitinde saptanan hata oranının %6,67 olduğu hesaplanmıştır. Geliştirilen sistemde araçların tespiti, anlık hız tespitinde hesaplanan başarı oranlarının %72,35 ile %100 arasında olduğu tespit edilmiştir.
In recent years, there has been a rapid increase in the number of registered vehicles worldwide. This increase, coupled with advancing vehicle technologies, has led to a significant increase in traffic congestion and accident risk. Various efforts have been made in many countries to reduce the risk of accidents. On 1 January 2004, the European Conference of Ministers of Transport (ECMT) was established to conduct research on road transport and connectivity. The Speed Management report, jointly produced by the World Health Organization (WHO) and the ECMT in 2006, is one of the most important studies in this field. A prevalent consensus drawn from speed and accident prevention research in numerous European countries underscores the necessity for augmenting the incorporation of technology within enforcement systems. The use of electronic systems for road enforcement would provide a more practical and cost-effective way to reduce the use of field staff and vehicles. Due to this factor, many countries have widely embraced the utilization of diverse electronic monitoring systems. Through these systems, controls such as speed limit violations, lane violations and seat belt usage can be realized. In Türkiye, speed limits in electronic monitoring systems vary according to the type of highway and vehicle used. These systems include speed limits specifically set for different road types and vehicle categories. The speed estimation process based on the time of passing vehicles over a given distance was tested using a Kalman filter. In the control process, each vehicle was identified with a unique ID value and its unique ID number, average speed and instantly captured snapshots were recorded in case of a speed limit violation. In a vehicle speed detection example taken from the tests, it was determined that the speed calculated with the Kalman filter was 70.27 km/ h, the speed calculated with the Odorless Kalman filter was 69.99 km/ h, and the speed calculated with the Multivariate Kalman Filter was 70.02, while it showed a speed value of 70.34 km/ h in a pass at a constant speed of 70 km/ h. As a result of the constant speed passes with the test vehicle, it was calculated that the average error rate in determining the vehicle speed was 6.67%. It was determined that the success rates calculated in the detection, speed determination and counting of vehicles in the developed system were between 72.35% and 100%.
In recent years, there has been a rapid increase in the number of registered vehicles worldwide. This increase, coupled with advancing vehicle technologies, has led to a significant increase in traffic congestion and accident risk. Various efforts have been made in many countries to reduce the risk of accidents. On 1 January 2004, the European Conference of Ministers of Transport (ECMT) was established to conduct research on road transport and connectivity. The Speed Management report, jointly produced by the World Health Organization (WHO) and the ECMT in 2006, is one of the most important studies in this field. A prevalent consensus drawn from speed and accident prevention research in numerous European countries underscores the necessity for augmenting the incorporation of technology within enforcement systems. The use of electronic systems for road enforcement would provide a more practical and cost-effective way to reduce the use of field staff and vehicles. Due to this factor, many countries have widely embraced the utilization of diverse electronic monitoring systems. Through these systems, controls such as speed limit violations, lane violations and seat belt usage can be realized. In Türkiye, speed limits in electronic monitoring systems vary according to the type of highway and vehicle used. These systems include speed limits specifically set for different road types and vehicle categories. The speed estimation process based on the time of passing vehicles over a given distance was tested using a Kalman filter. In the control process, each vehicle was identified with a unique ID value and its unique ID number, average speed and instantly captured snapshots were recorded in case of a speed limit violation. In a vehicle speed detection example taken from the tests, it was determined that the speed calculated with the Kalman filter was 70.27 km/ h, the speed calculated with the Odorless Kalman filter was 69.99 km/ h, and the speed calculated with the Multivariate Kalman Filter was 70.02, while it showed a speed value of 70.34 km/ h in a pass at a constant speed of 70 km/ h. As a result of the constant speed passes with the test vehicle, it was calculated that the average error rate in determining the vehicle speed was 6.67%. It was determined that the success rates calculated in the detection, speed determination and counting of vehicles in the developed system were between 72.35% and 100%.