Road Traffic Decision Support System Road traffic management has long been a complex issue and seems

likely to continue to be so. Road traffic laws and policies depend on a large number of factors. Making a correct decision for traffic management can be difficult because decision-makers need to analyze and absorb a large quantity of information. Therefore, there is a need for a better control and a reliable and consistent system to help simplify the traffic decision making process. By incorporating the variables involved in traffic management such as the numbers of accidents, traffic violations, and traffic policemen on duty into an artificial intelligence technique, it is possible to build a traffic decision making system to help decision makers for analysis of traffic laws and policies. Fuzzy logic allows us to provide recommendations on whether traffic decision makers should, for example, give priority to increasing the number of traffic policemen on duty, increasing fines, or installing better/more safety devices. The variables involved in traffic management are dynamic and complex: traffic accidents, traffic violations, available traffic policemen, available traffic cameras, etc. There dynamic variables are fuzzified and used to be analyzed with a fuzzy inference system. A fuzzy rules based system is developed using the expert knowledge to generate effective policies. The system priorities the possible traffic policies. They include if a given city should increase the number of traffic policemen on duty, increase fines, impose stricter speed limits, install better safety devices, or provide better education for drivers. The system itself would not decide the policies, but would provide traffic decision makers with better information so they could make better decisions. The first and probably most important task in the process of designing a fuzzy logic decision support system is the identification of influencing parameters and to their relation on decision making process. In a traffic decision-making system, the traffic factors contributing primarily to a decision making and the typical traffic decisions are the fuzzy variables. Five influencing factors and four typical decisions were determined. Influencing factors: The first and most important factor is the number of traffic violations. A large number of traffic violations would suggest that the traffic system was out of control, either because of a shortage of available personnel and equipment such as traffic policemen or traffic cameras, or because that the

punishments and fines for having broken the law is not acting as a deterrent to drivers. The number of traffic accidents is a second important factor influencing the making of most traffic decisions making. It is a fact that reducing the number of traffic accidents is the principal goal of all traffic managements. A good traffic system could be seen as one that has a low rate of traffic accidents. The following two factors are very similar and so are mentioned together. These are the number of vehicles per traffic policeman on duty and the number of vehicles per traffic camera. These two parameters are determined as influencing factors because they are considered as a measure of controlling and monitoring traffic. Thus, to keep the traffic controlled, the number of vehicles per policeman and the number of vehicles per traffic camera should not be high. The values of these two factors in a particular period are based on how many vehicles are used at that period, and also on how many traffic policemen and traffic cameras are available at that period. The rate of traffic jams is the last and least important factor because it does not directly affect traffic decisions, but is combined with the accident factor. For example, the decision that might be made when the number of accidents is large and the rate of traffic jams is low, is not the same as the decision that might be made when the number of accidents is unchanged but the rate of traffic jams is high, because the high rate of traffic might cause many minor accidents. Typical Decisions: There are a number of typical decisions made by traffic policy maker based on the identified influencing parameters. Launching a traffic awareness campaign is the first typical decision. The wide concept underpinning traffic awareness campaigns is creating awareness about motorist and pedestrian safety, in order to minimize or prevent traffic accidents. The number of accidents, the number of traffic violations, and the rate of traffic jams are the three factors that influence this decision, but the number of accidents is given a larger weighting. The second typical decision is increasing punishments or fines. Sometimes, punishments or fines of breaking the law are not acting as a deterrent to drivers. Hence, increasing the punishments or fines is considered as an applicable solution to reduce the number of traffic violations. This decision is influenced by three factors; the first factor is the number of violations,

which has a large weighting. The second and third factors are the number of vehicles per traffic policeman and the number of vehicles per traffic camera. The third and fourth typical decisions are increasing the number of traffic policemen on duty and increasing the number of traffic cameras. These two decisions are made to reduce the number of violations, when the number of vehicles per traffic policeman is high or the number of vehicles per traffic camera is high. Table I summarizes the relationships between the influence factors and the typical decisions
Table 1 : The Influence Factors and the Typical Decisions Typical Decisions Influencing Factors Number of traffic accidents Launch traffic awareness campaign Number of traffic violations Rate of traffic jam Number of traffic violations Increase punishment or fine Number of vehicles per traffic policeman Number of vehicles per traffic camera Increase the number of traffic policeman Number of vehicles per traffic policeman Number of traffic violations Increase the number of traffic camera Number of vehicles per traffic camera Number of traffic violations

Fuzzy Logic in Decision Making: After identifying all parameters (input and outputs) needed for building our fuzzy model, fuzzifying those parameters was the next stage. For each of those parameters, appropriate linguistic variables associated with their numerical ranges have been specified. Then we have used linear triangular membership functions to design the fuzzy sets for all linguistic variables. Figure 1 and 2 shows the membership function of the input functions and output functions.
Very Low 1.0 0.5 0.5 Very Low Very Low 1.0 1.0 0.5 0.5 Low Low Medium Medium High High Very High Very High Low Medium High Very High Very Low 1.0 Low Medium High Very High

1000

3500

5000

7500

10000

100

350

500

750

1000 High 7500 7500 Very High 10000 10000

Traffic Violation

Very Low 1.0

Traffic Low
1000 1000

Accident Medium
3500 3500 5000 5000

Vehicle per cameras Vehicle per policeman

10

35

50

75

100

Rate of traffic jam

Figure 1: The triangle membership for the various input function

0.5

0.5

10

10

Launch traffic awareness campaign decision

Increase Punishment/Fine decision

Medium 1.0

High

Very High

0.5

0.5

Increase the number of policeman decision ddecision

10

10

Increase the number of cameras decision

Figure 2: The triangle membership for the various output function The next step was building the rule-based structure. This was done by breaking the control problem down into a series of IF X AND Y THEN Z rules that define the desired system output response for given system input conditions. Table 2 shows a selection of these rules.

Table 2 : Selective Rules-based structure IF Violations is medium AND Jam is low AND Accidents is very high Violations is high AND Vs/Policeman is high AND Vs/Camera is high Violations is very high AND Vs/Policeman is high AND Vs/Camera is very low Violations is very low AND Vs/Policeman is high AND Vs/Camera is very high Then Launch Traffic Awareness Campaign is High Increase Punishment/Fine is Medium Increase Policeman is Very High AND Increase Camera is Low Increase Policeman is Very Low AND Increase Camera is Very Low

The final stage of designing our traffic fuzzy logic system was applying an appropriate defuzzification method in order to obtain a crisp value (decision priority) for each decision. The Centroid method was used to calculate the approximate centre of gravity of the distribution for the fuzzy sets. The whole structure of the traffic fuzzy inference system, consisting of five inputs parameters that map to four outputs, which have been translated into fuzzy values (fuzzy sets). In conclusion, the proposed system has shown that Fuzzy logic based approach has a considerable potential to be used in the development of a road traffic management system to support decision-takers. However, only the technical feasibility of the system with the limited influencing variables/ factors which directly affect traffic decisions was demonstrated. The drawback of the above proposed system was that it did not take in all the factors that may influence the traffic decision-making process such as road safety, drivers behaviors, the weather condition, holiday seasons, etc., All these factors should be taken into consideration.