Distance Detection on an Autonomous Vehicle Using the Triangle Similarity Approach

Currently, autonomous vehicle advancement is very rapid, especially for developing algorithms or methods to support autonomous systems. Autonomous vehicles have become a hot topic in intelligent transportation and will improve safety on the road. This research aims to propose an autonomous vehicle safety system by detecting the distance between vehicles as initial research for the concept of autonomous vehicles regarding safe driving. For the car detection process, the Histogram of Oriented Gradients (HOG) method is used to obtain feature extraction from car objects. In the classification process, Support Vector Machine (SVM) linear classification is used to classify the car object. The main contribution of this research is to propose vehicle distance detection in autonomous vehicles using a triangle similarity approach. The triangle similarity is used to detect and calculate the distance (focal length) between the camera and the detected object. The decision boundary is obtained regarding the safe distance to the vehicle by marking the box in red for unsafe distances and green for safe distances. The result showed that accuracy reaches 75%, the MSE for bounding box has a value of 0.2, and the MSE for distance status has a value of 0.125.