System concept of a sensor network for vehicular traffic monitoring based on acceleration sensors

Rivas Rodriguez, Antonio Javier; Heinen, Stefan (Thesis advisor); Mokwa, Wilfried (Thesis advisor)

Aachen (2018) [Dissertation / PhD Thesis]

Page(s): 1 Online-Ressource (xxi, 142, cii Seiten) : Illustrationen


Efficient usage of the road infrastructure is crucial for the progress of modern cities. Advantages of a good understanding of the traffic behavior are reduced traveling times, identification of conflict zones, optimization of traffic lights, prevention of accidents, among others. Moreover, it is possible to determine if the current infrastructure can meet the requirements of mobility for persons and goods or if new roads are required. Traditional traffic detectors like video-cameras, radars, or inductor-loops provide limited information about the traffic conditions, because they are installed on fixed zones on the roads being not possible to know if a driver was behaving correctly before approaching to their measuring range, or if an accident has occurred beyond their range. Moreover, they are affected by the ambient conditions like light, noise, fog, rain. Therefore, to get better information of the situation on the roads that helps to increase the efficiency of the current infrastructure a new approach is necessary. One approach is to interconnect vehicles with the infrastructure to continuously interchange information of the traffic situation. Many automakers and other companies are working towards Car-to-X technologies, where the X means cars or infrastructure. Such technologies allow to communicate cars with the infrastructure providing real-time information about obstacles, accidents, traffic signs, navigation, location of places, among others. As a result, the number of accidents are reduced, also less time is spent on the roads reducing the contaminants emission. The technology underlying the Car-to-X communication is based on sensor networks. Sensor networks consist of individual nodes each one including one or more transducers (combination of sensors and actuators) to interact with their environment, a power source, a control unit, and a wireless transmitter. Each node can operate independently of each other and can communicate with its neighboring nodes. Compared with traditional detectors they are less susceptible to the weather conditions and allow an uninterrupted monitoring of the vehicular traffic at all points on the road. Given that nodes communicate to each other, they can transmit information of the traffic situation to nodes far away and anticipate to the upcoming traffic events, allowing to adjust the traffic lights to make more fluid the traffic on the roads and to prevent accidents. Moreover, they potentially can track individual vehicles along the roads, so it is possible to know how drivers behave all time. Driving assistance systems implemented on modern vehicles can be enhanced by interchanging information with net of sensors placed on the roads. Advantages of such interconnectivity are reception of traffic information in real-time, or complexity reduction in the driving assistance systems by letting specific tasks to be performed by the infrastructure. As a result, these assistance systems would be more affordable contributing to a faster adoption in cheaper vehicles, which in turn will contribute to make the roads safer due to all the previously mentioned advantages. An interesting proposal about intelligent roads is presented in the article “Intelligent Road Infrastructure - A Concept Study”. It proposes the use of a two-dimensional net of MEMS accelerometers placed on the width and length of the road's surface to monitor in real-time the vehicular traffic conditions by measuring the vibrations produced by the vehicles on the roads. Motivated by that proposal, this dissertation presents a concept study of a one-dimensional, roadside vehicular traffic detector based on the working principles explained in that article, capable of determining the presence, travel direction, speed and type of vehicle passing on the roads. For the presented study, high-sensitive piezoelectric acceleration sensors based on the same measuring principle as MEMS accelerometers are used. The data obtained with these sensors is used to study with high detail the amplitude and frequency range of the road vibrations to develop the algorithms to calculate the presence, travel direction and speed of the detected vehicles. The results obtained with the developed system concept are going to define the specifications for future specific-purpose MEMS devices that will be integrated in a closer version to the original idea of the traffic detector in which a 2-D net of MEMS sensors is to be implemented. Commercial MEMS accelerometers were not considered for the study, because they have sensitivities of tens of milli-g, which is the upper limit of the estimated vibrations amplitudes on the roads. The accuracy obtained with the developed system is acceptable, having on average 80% for vehicles detection, which can be increased up to 90% by discarding the bicycles. The accuracy in the determination of the travel direction was on average 90%. The calculated speeds have an error average of 27% with respect reference measurement. The developed algorithms prove that monitoring the traffic flow using only the road vibrations is possible. Advantages of the developed system against traditional traffic detectors like video-cameras or radars is that it is less affected by the weather conditions like wind snow or the night. Moreover, its portability makes it easy to relocate in different roads to monitor the traffic. This work sums to a continuously growing list of works aimed to add intelligence to the roads infrastructure by using sensor networks and complementing in-vehicle assistance systems with the goal to improve the efficiency and safety on the roads.


  • REPORT NUMBER: RWTH-2018-220834