• Lukás Stranovsky
Car accidents are one of the major causes of death in the world today.
Studies proved that the main causes of car accidents are distracted drivers, overspeeding and driving under the influence of alcohol or drugs. According to NSC (2016) and NHTSA (2018), most accidents on roads are caused by human factor and solution to this are self driving autonomous vehicles. However, this is not yet possible because of reasons such as Infrastructure and road conditions, People do not trust machines, Accountability in case of an accident is not clear, Technology is not up to the task.
This report study possibilities of technology - Light Detection and Ranging (LIDAR) - a sensor that is mounted on self-driving vehicles to detect obstacles.
We propose a system that is capable of detecting other cars on the road consisting of 4 modules: Preprocessing, Ground removal (RANSAC), Clustering (DBSCAN) and Classification (SVM). Furthermore, an experiment is carried out on public domain benchmark dataset KITTI. In the experiment, 6 different models are trained. The results showed excellent accuracy in terms of detecting cars.
Publication date31 Oct 2019
Number of pages78


ID: 313538523