Vanishing point detection for self-driving car using harmony search algorithm

Abstract

Self-driving or autonomous vehicles require an ability to detect road lanes. In order to do so, a vanishing point should be first detected because the vanishing point exists on the extended lines of road lanes. For detecting the vanishing point, a random sample consensus (RANSAC) algorithm has been generally utilized. However, the performance of RANSAC is sometimes not so good but fluctuated. Thus, this study proposes a new approach to estimate the vanishing point using a harmony search (HS) algorithm. Results show that HS stably estimates vanishing points with respect to statistics when compared with RANSAC. We hope this model to be utilized in self-driving car in the future.