Dynamic Control Strategy

Description: This research develops a control strategy that takes into account the available resources, location of shelters, accessible data, and the network clearance time. As a result, this approach utilizes officers for guiding traffic because they are an available in any city, they can provide information using mobile devices, and their presence might calm drivers who are more likely to follow orders from policeman than any traffic signal. The proposed technique is based on an algorithm that utilizes a predictive application for estimating the state of traffic. Thus, it also employs a heuristic to evaluate if it is worthy or not to reroute an evacuation exit. To evaluate the new control strategy, a traffic simulator was created. Its implementation is based on the parallel computing to overcome the computational cost of the different modules of the software application. Thus, it provides metrics such as total evacuation time, number of evacuees at shelter for every time step, and the state of intersections during the evacuation. The model was tested in two environments with different distribution of vehicules. The outcomes where compared with those obtained from an uncontrolled environment and those from the random approach. According the metrics, the dynamic strategy has the best performance because its traffic distribution is better and its average evacuation time is lower than of the uncontrolled scenario.

Topic: TRAFFIC MANAGEMENT
ARTIFICIAL INTELLIGENCE
KALMAN FILTER
MACHINE LEARNING

Get full text: http://repositorio.educacionsuperior.gob.ec//handle/28000/1606