Intelligent Traffic Control and Pedestrian Crossing System
C.H.W. Jayawardane
Faculty of Information Technology, University of Moratuwa, Sri Lanka
Abstract - This paper proposes an intelligent traffic light and pedestrian crossing system that can be used to address traffic congestion in a four way junction. The approach is primarily focused on creating a system which dynamically responds to traffic conditions and road users. The system makes use of sensors to detect the presence of a vehicle. A number of these sensors placed accordingly are used to identify the number of vehicles in a particular lane. Based on this input, which is the vehicle line length in a lane the traffic control algorithm implemented in our system takes the necessary decisions. This algorithm determines the lane that should be given the most priority and up to what amount of time. This feature allows maximum efficiency in controlling the flow of traffic. The pedestrian crossing system uses image processing technology to identify the number of waiting pedestrians. When this number reaches a defined value they are given the chance to cross the road. Else after a maximum time limit is exceeded and the required number has not yet gathered the remaining will be given a chance to navigate safely across the road. A count down timer display is used to show the remaining time the pedestrians have. It can be illustrated that the proposed system provides a cost efficient and environmentally friendly method to all road users.
I

the day to reflect rush hours and normal time, the lights do not have the ability to adjust their own sequence according to varying traffic conditions [1]. Many traffic light systems operate on a timing mechanism that changes the lights after a given interval. An intelligent traffic light system senses the presence or absence of vehicles and reacts accordingly. The idea behind intelligent traffic systems is that drivers will not spend unnecessary time waiting for the traffic lights to change [2]. This project focuses on implementing a system where the length of traffic in a lane is detected using sensors. Based on the sensor inputs given, a decision will be made by the processor according to an algorithm, whether to continue traffic flow on the current lane or to allow the green light to a different lane. The pedestrian crossing system uses image processing technology. Instead of allowing a predefined time period within each cycle time for pedestrians to cross the road, a video camera placed accordingly will count the number of pedestrians waiting to cross the road. After it reaches a given number the system will interrupt the traffic flow and allow the pedestrians to cross the road. The upcoming sections of the dissertation are organized as follows. The second section describes the work carried out by others related to this field. The third section focuses on describing the types of technologies that has been used. The approach we have taken to develop this system is discussed in section four. The designs of each module and their interactions are discussed in chapter five, analysis and design. Section six has been dedicated to how hardware and software have been integrated to implement the system. The final two chapters evaluate the system and identify the conclusions. II LITERATURE REVIEW

INTRODUCTION

As Sri Lanka is still a developing country, the transport system currently in use today is not quite up to the standards to meet the growing needs of the society. When the current transportation system is not able to sustain the required efficiency it will result in major traffic problems, accidents and pollution. The delay getting the required raw material into a factory will affect not only the factory but the market too. When an individual is late for work certain decisions and work may get delayed and it may cripple the work of the company. We can see that a single delay in a traffic jam can lead to a chain of events that will directly affect the country’s economy. Road safety has also taken a back seat in recent times. Pedestrians are either unaware or ignorant of the rules governing them on the road. Most of the time pedestrians do not use the pedestrian crossing but rather try to navigate through the oncoming traffic before their turn and the chances of a road accident is very high. These issues are due to the growing numbers of road users and the limited resources provided by our infrastructure to support them. At present, traffic control systems follow hard-coded sequences for directing traffic. Although these patterns may change several times during

A broad range of diverse technologies, known collectively as Intelligent Transportation Systems (ITS) hold the answer to many of the transportation problems that are being faced today. These technologies can be broadly categorized into the following. Inductive loops are one such example. The inductive loops can be placed in a roadbed to detect vehicles as they pass over the loop. The vehicle is identified by measuring the vehicle's magnetic field. The simplest detectors simply count the number of

vehicles during a unit of time, while more sophisticated sensors estimate the speed, length, and weight of vehicles and the distance between them. These types of loops can be placed in a single lane or across multiple lanes, and they work well with very slow or stationary vehicles as well as vehicles moving at high- speeds [3]. However maintenance on buried active inductive loops is difficult and expensive. They can also be unreliable because such loops are often prone to breakdowns, particularly in hostile weather environments [4]. Sensing systems are vehicle and infrastructure based networked systems that are installed or embedded on the road, or surrounding buildings. Examples for these types of sensors are reflective and infrared sensors. They react to motion and send signals to trigger the light changes. Traffic flow measurement and automatic incident detection using a video cameras is another form of vehicle detection. Video from black and white or color cameras is fed into processors that analyze the changing characteristics of the video image as vehicles pass. The cameras are typically mounted on poles or structures above or adjacent to the roadway. Most video detection systems require some initial configuration. This usually involves inputting known measurements such as the distance between lane lines or the height of the camera above the roadway [3]. However the inability of this technology to detect stationary objects and slow velocity objects exist [5]. Although many research projects have successfully been carried out related to this field across the globe and developed technologies have been efficiently implemented, Sri Lanka is yet to benefit from advanced traffic light systems since our current infrastructure cannot yet support such sophisticated systems. We are still using transitional timed traffic light systems and other methods such as fly over bridges to minimize the traffic congestion. The Arthur C. Clarke Institute for Modern Technologies (ACCIMT) has taken a first step to change this situation, designing the ACCIMT Traffic Lights System. This is a microcontroller-based flexible system with 16 programmable traffic pattern combinations. The userfriendly interface enables, scheduling at least 08 different timing programs in a day with separate green time setting for weekdays, Saturdays, Sundays and other holidays. The system can be reconfigured on-site, without interrupting the operation of traffic lights [1]. This system does not fully satisfy the criteria of an intelligent system. However, the most resent project of designing the overhead bridge at Nugegoda has taken the first step to introduce this technology to Sri Lanka. The ACCIMT had developed a system where the roads with heavy traffic at any given time can be given priority, irrespective of the original program, by installing certain sensors [4].

TABLE 1 Comparison of current systems implemented in Sri Lanka with the proposed solution The current system in Sri Lanka Special lights for pedestrians to cross at junctions. These lights have a system that indicates how long it will take for the colors to change by giving a countdown. Certain systems are installed with cameras. For instance, Horton Place junctions have cameras. Average timings for changing of lights at each junction are decided by the data collected through a manual count. Proposed Solution The proposed solution takes into account the amount of traffic in each direction. And dynamically changes the signal output. The pedestrian crossing system counts the number of pedestrians that are waiting to cross the road and allocates time accordingly.

III Image processing

TECHNOLOGY ADAPTED

Identifying human faces in an arbitrary image is a fundamental step in video monitoring. Face detection algorithms have been widely studied in several disciplines, such as image processing, computer vision, and neural network. For the purpose of this project image processing was used and an appearance-based method has been used [8]. An appearance based method uses features of a human face such as the length between the eyes, length of eye to nose to identify a face [9]. Image processing was used to implement the pedestrian crossing system. This technology was mainly used as an alternative to sensors. By using image processing we were able to replace a large number of sensors to identify a human. The sensors had no accurate method to justify a human face. The risk of the system detecting a false input was therefore high. With the use of this technology this issue could be overcome. Therefore image processing technology was the best solution for the problem. Microcontroller A microcontroller is an integrated chip that is a part of an embedded system. The microcontroller includes a CPU, RAM, ROM, I/O ports, and timers. But they are designed to execute only a single specific task or to control a single system, they are much smaller and simplified so that they can include all the functions required on a single chip [6]. The system is designed using the PIC 16F877A microcontroller. Fig. 1 represents its pin diagram. A microcontroller contains general purpose input/output pins. They are software configurable to either an input or an output state. A micro controller was used for this project due to the following reasons.

Fig. 1 PIC 16F877A

Each lane has three sensors placed at a distance. Using these, the length of the vehicle line can be calculated. Based on these lengths the priority that each line is given and the amount of time the green light is given is determined. If two or more lanes are having the same vehicle line length the selection will be decided through another parameter. The amount of time lapsed after the last transition from green to red signals. This selection and timing control will be done by the computer based on an algorithm. The output from the parallel port from the computer is sent to the micro-controller through a level converter. The level converter is used to magnify the signal. Each LED is connected separately to the microcontroller and it is programmed based on the outputs by the microcontroller to change the lights in each lane accordingly. The video camera is placed on the side of the road in a position that can cover the entire area that the pedestrians can gather. Still images are taken by this camera with a five second interval. These images are inputs to the computer program. They are analyzed using the appearance based method and passed on to a Dynamic Link Library (DLL) function. The function returns the number of faces waiting to cross the road. This number is constantly compared with the predefined number in the program till it reaches the defined number. When the value returned by the function is equal to or more than the defined number the traffic flow will be disrupted by generating a signal to the microcontroller and the pedestrians will be allowed to cross the road. This is displayed by invoking a countdown timer display interface. The output is the remaining time available for the pedestrians. In a normal situation where the traffic flow is very light and if only one or two pedestrians are waiting to cross the road a maximum time period is defined and a signal will be automatically generated to the microcontroller to halt the traffic flow and to allow the pedestrians to cross the road.

Size and Weight: Microcontrollers are compact and light compared to computers. Reliability: Since the architecture is much simpler than a computer it is less likely to fail. Speed: All the components on the microcontroller are located on a single piece of silicon. Hence, the applications run much faster than it does on a computer [7]. Input switch and denouncing circuit An input switch is a simple circuit that can be used to obtain an input to the system. However when the switch is triggered spikes of low and high voltages will occur resulting in a series of high and low signals. These spikes can be interpreted by a digital circuit as more than one pulse instead of one clean pulse or transition from one logic state to another. As a solution to this problem an analog debouching circuit is used. IV APPROACH

The system introduced in his paper is focused on two main user groups, drivers of vehicles and pedestrians. Fig. 2 represents the approach to developing the proposed system. When a vehicle is parked over or crosses the input switch an electric pulse will be created. This pulse will be input to the computer through a micro-controller using a serial port. When the switch is triggered spikes of low and high voltages will occur across that switch resulting in a series of high and low signals. These can be interpreted by a digital circuit as more than one pulse instead of one clean pulse or transition from one logic state to another. This error will result in the system identifying one vehicle as many. As a solution to this problem an analog debouching circuit is used. The analog solution relies principally on a capacitor, which plays the role of resisting the voltage changes on the output. In other words, this will prevent the output to change too fast which will prevent high and low pulses to appear on the output.

Fig. 2 Co-approach to hardware and software design

V

ANALISIS AND DESIGN

A thorough analysis of the system as enabled the solution to be structured as follows. Fig. 3 illustrates a high level design diagram of the intelligent traffic light and pedestrian crossing system. This represents the main modules and the interaction between them. Traffic Control System The traffic control system consists of three main modules. The first is the detector. The detector embedded into the road is able to sense the presence of a vehicle as it passes over or stops on top of it. The electronic signal generated is then carried to the control unit through a serial port. The second module is the controller, which represents the brain of the traffic control system. It consists of a computer and a microcontroller that controls the selection and timing of traffic movements in accordance to the varying demands of traffic signal as registered to the controller unit by the detectors, based on an algorithm. The third module is the signal visualization or signal face. It comprises of three Light Emitting Diode (LED) s red, green and yellow which lights up and directs the traffic flow intelligently according to the output from the control unit.

The first is the detector. Here a video camera placed on the side of the road detects the number of pedestrians that are waiting to cross the road. The second module the controller consists of a computer and a program that takes a decision as to whether the pedestrians are allowed to cross or not. The decision is based on two parameters. The number of pedestrians and the amount of time elapsed from the previous time of crossing. The third module is the signal face. If the controller allows the pedestrians to cross the road a countdown timer is displayed on the monitor to show the time remaining to cross the road else a “Please Wait” sign will be displayed. VI Traffic Control System In each lane, three push switches have been installed. Therefore, the program will check first the condition of the sensors, whether they are triggered or not. The total number of sensors triggered will be used in to calculate the length of the vehicle line and appropriate timing for the green signal to illuminate. After the green signal finishes the illumination timing, the yellow signal will illuminate for two seconds and then finally the red signal will illuminate. Afterwards it will continue in a normal clockwise sequence until another lane satisfies the condition of maximum vehicle line length. This will disturb the normal flow and give priority to that lane. A single three lamp traffic light is considered as a finite state machine. It has three states, Red, Yellow, and Green, which are the outputs. A single input for the traffic light is defined, with values 0 for no change and 1 for change. The traffic light is installed in pairs, with two pairs per intersection. This is done to minimize the output. As the output of one will be the inverse of the other. Therefore, in the system one pair of lights was used to control traffic in the north-south direction, while the other pair controls the east-west direction. The two pairs of lights must be synchronized to avoid a contradictory signal combination that could produce fatal results. Since the lights that make up a pair mirror each other, they are considered as a single light. The different outputs are grouped to a single unit. Each combined output describes the color of the northsouth light along with the color of the east-west light. Pedestrian Crossing System Fig. 5 represents the implementation steps of the pedestrian crossing system. The camera operates in an infinite loop. At every five seconds interval an image is taken. This image is analyzed to get the number of faces. IMPLEMENTATION

Fig. 3 Top level design of the proposed system

Pedestrian Crossing System The pedestrian crossing system also consists of three modules.

destination of the driver. The time spent on getting to the destination previously and after the implementation of the prototype system. A similar set of questions will be given to the pedestrians. The gathered data will be analyzed using statistical techniques. The numerical values will be presented with the use of relevant graphs and charts. VIII CONCLUSION

Fig. 5 Pedestrian crossing system

VII

EVALUATION

A formal evaluation of the proposed system has not yet been finalized. However the components of the system have been evaluated separately so as validate their performance. The push button switch embedded in the road was tested using model vehicles. The main aim was to test their sensitivity to the weight of the vehicles. The performance of the traffic light controlling algorithm was examined with the use of a simulator. The inputs were given through the key board in the form of an eight bit sequence. The color light changes corresponding to each sequence was viewed through the simulator. This method of evaluation helped to identify the deadlock situations. The signal transmission circuit was identified as functional, though the accuracy of the output LED s. The pedestrian crossing system was tested using a web cam. The web cam was used to get an image of the surrounding. The accuracy of the program was tested by setting the predefined value to one. If more than one face was detected the countdown timer would be initialized to twenty seconds and the countdown would begin. A formal evaluation strategy has been proposed so as to validate the system performance. The strategy has two main phases. Phase one is focused on testing the system through simulation. This method requires no human participation and is used to gain a basic. The second phase involves gathering of actual data through a questionnaire. This testing is done in a real life environment. The prototype system will be implemented in a four way junction. A sample of pedestrians and drivers that are regular users of the route are identified and used as the participants of the test. A short questionnaire will be used so as to motivate the participants to give a feedback. The questions will be focused on getting the details of the

This paper proposes a sensor based application coupled with image processing to develop a traffic algorithm that is dynamic. According to the literature review the methods and the technologies used to design the system are below the current standards of the world. But the main objective was to design a system that could be implemented in Sri Lanka which is supported by the current infrastructure and the technology available in the country. Sensor technology is low in cost compared to inductive loop and video detection technology. Therefore it has been chosen for to implement the system. In practice most traffic lights are controlled by fixedcycle controllers. A cycle of configurations is defined in which all traffic gets a green light at some point. The split time determines for how long the lights should stay in each state. Busy roads can get preference by adjusting the split time. The cycle time is the duration of a complete cycle. Fixed controllers have to be adapted to the specific situation to perform well. Setting the control parameters for fixed controllers requires enormous amounts of data to be gathered and analyzed. These controllers have to be updated regularly due to seasonal changes in traffic situation. The intelligent traffic light system presents several advantages. Since the waiting time of the vehicles for the lights to change is minimal, the emission of harmful gases such as carbon dioxide from the vehicles is reduced. This will give a positive effect on our environment by reducing the greenhouse effect in our atmosphere. The system will also save the motorists time and reduces their frustration while waiting for the lights to change since it helps reducing congestion in the traffic intersections. By considering the above we can come to the conclusion that the objectives of the project have been met successfully. However several minor limitations can be identified in the system. The application requires that all road uses such as drivers and pedestrians to be well disciplined and that they abide by all road rules and regulations. This is needed to ensure that all road users await their proper turn. The system has also not taken into consideration how severe weather conditions can affect the performance of the system. For example on a rainy day it can be quite difficult to analyze the features of a face through a blurred image.

To take a step further to improve the current system, a solution needs to be found to direct the traffic at the time of an emergency such as an accident. The current system is not intelligent enough to handle such a situation. Therefore an application can be proposed to send a message to the nearest police station. This can be built on top of the existing system. REFERENCES
[1] Tzafestas et al, (1999). Advances in Intelligent Autonomous Systems: Microprocessor based and Intelligent Systems Engineering, Kulwer Academic Publisher. [2] Intelligent transportation system - Wikipedia, the free encyclopedia URL: http://en.wikipedia.org/wiki/Intelligent_transportation_system [3] Sun W. and K.C. Mouskos, (2000), Network-wide traffic responsive signal control in urban environments, Proc. 4th Intl. Conference, Computational Intelligence & Neurosciences, and Atlantic City, New Jersey. [4] Magnetic sensor system and method for installing magnetic sensors URL: http://www.freepatentsonline.com/7005850.html [5] Jung Y K., Ho Y S. (1999) Traffic parameter extraction using videobased vehicle tracking, Proceedings of IEEE / IEEJ / JSAI International Conference on Intelligent Transportation Systems. [6] What is a Microcontroller? URL:http://www.wisegeek.com/what-is-a-microcontroller.htm [7] PIC16F877 Microcontroller Tutorial rev. A URL:http://www.pages.drexel.edu/~cy56/PIC.htm [8] M.-H. Yang et al, “Detecting Faces in Images: A Survey”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.24, No. 1, Jan. 2002. [9] Yulei Weng and Alex Doboli, Smart Sensor Architecture Customized for Image Processing Applications, State University of New York at Stony Brook, Stony Brook, NY 11794