Smart Traffic Lights

Abstract

The problem that exists in our society is lack of efficient traffic control system. As a result of this, traffic congestions take place on the roads and fuel is wasted and the time of common man is also wasted as they have to wait for a long period of time in traffic jams. Also in the traffic jams, there are a lot of ambulances and they carry injured patients and if time exceeds, then the patients may pass away. Another problem is that a lot of poisonous gases are released that harm the environment such as carbon dioxide, carbon monoxide. So in the light of these problems faced by our society we have proposed smart traffic lights. One of the biggest problems faced by us was that there is very little material available on internet regarding an intelligent traffic control. Another challenge is that the research papers are quite expensive so we do not have access to them. So we have to look for information from somewhere else. Another challenge is that the group members have different sections, so this means that there is conflict in schedule. This makes it hard to coordinate. This creates problem and therefore group meetings are inadequate. But the biggest challenge for us is to make a working product in such a limited time. We have to make certain choices because of limited time. Another challenge is that we are required to make a very advanced system with quite less experience. There are different approaches that have been proposed to solve the problem of traffic congestion. One of these approaches is reinforcement learning. In this approach, the basic idea is that a car is considered to be an entity and it makes its decision in such a way that there should be an increase in the benefit. The benefit keeps adding up as the entity continues to make its decisions. After each adjustment move of the entity in its habitat, the value of the reward changes. A multi-layered perceptron is taught to forecast the size of row for succeeding unit of time and adequate forecasting can be done. Each entity has two different types of control i.e. simple control and elaborate type of control. A machine is placed on the road that calculates the number of cars on the road and transmits this information to a server which then performs forecasting of traffic congestion. It is possible to employ Markov decision problems for a scenario of an entity communicating with its habitat. Another approach is genetic algorithm. In this approach, the path of all the cars is encoded to a chromosome and “path shortening” is a feature which is applied in order to improve “hunt” performance. This model involves making a few assumptions such as the place of going off, the place of disembarkation and the time of going off is known to the system. Those cars which have reached their destinations and those cars which have not yet started to move do not affect in predicting the traffic. The road network is considered to be a graph that has edges and nodes. All the edges are “cut” into blocks and the velocity is calculated for each block. The factor which is used to calculate the performance is called Our proposed solution is traffic control system using pheromone model. According to this model, there is an entity called road unit. Each road unit has a sensor, a road agent and a server computer. The road sensor counts the number of cars and passes this information to road agent that basically calculates current traffic situation and send its output to server computer that performs forecasting. In this model there are five basic functions called current traffic situation, traffic density pheromones, congestion diffusion pheromones, evaporation rate and congestion forecasting pheromones. Our solution is better than existing systems called TCS (The Probe Car System) and VICS (Vehicle Information and Communication System) because our system is decentralized. In other words, it means that there is no single central server which is responsible for processing all of the information. This is beneficial because if the central server crashes, then the entire system will be unable to perform its required function and hence it will become useless. Another problem with central server is that if the number of requests keeps on increasing, then a point of time will arrive when the server will become bottleneck and the overall performance will be inadequate. These two are the main problems with having a single server and in our system there is no central server so our system will not have these problems. Other system that is more commonly used is the traffic lights that display the colors yellow, green and red for an equal duration. If we compare our system with this system, we see that our system is more advanced, has more features and functionality and is more beneficial. Another method to control the traffic is manually i.e. through traffic police. In comparison with our solution, this solution is quite ineffective and inefficient and our system is more convenient. Using Smart Traffic Lights road traffic will be managed in a systematic and well-planned manner. Our system is more responsive and accurate in forecasting traffic congestion than existing solutions. In the future, our system can be further improved by having a functionality that includes requiring customers to create their accounts and signing in to view the areas where they usually go to and saving the information of the locations where they go and automatically telling them the conditions of congestion in route to those areas and adding security measures against hackers with malicious intentions. Some of these security measures could include honey pot, firewalls, antivirus software, and information backup mechanism.