An Efficient Real Time Decision Making System for Autonomous Vehicle Using Timber Chased Wolf Optimization Based Ensemble Classifier

Abstract

An autonomous vehicle is anticipated to increase comfort, safety, energy efficiency, emissions reduction, and mobility. The development of autonomous vehicles depends on decision-making algorithms that can handle complex and dynamic urban intersections. Hence in this research, a TCWO based ensemble classifier-based instantaneous decision-making model involved in driverless vehicles is devised. The ensemble classifier is designed through the combination of the Timber chased wolf optimization (TCWO) for detecting traffic sign along with decision process is performed. The bidirectional long short term (BiLSTM) and convolutional neural network (CNN) is combined to create the hybrid ensemble classifier, which is more effective. The TCWO is developed by hybridizing the characteristics of GWO and COA that helps to optimizing the classifiers and boots the classification performance. The TCWO based GAN helps for finding lane in the data that effectively reduces the problems of misclassification by generating synthetic data and training the data to differentiate the original and the generated data. The TCWO-based ensemble classifier attained the values of 98.88%, 98.36%, 98.88% while detecting the traffic sign, and 2.41%, 2.41%, 7.39% while predicting the lane using TCWO-based GAN, which is significantly higher than the competent technique