The application of 2-dimensional LiDAR (Light Detection And Ranging) technology is sometimes constrained by the presence of data anomalies or noise that affects the accuracy in detecting real objects. If it is not handled properly, it can interfere with its work operations, especially if it is applied to autonomous electric vehicles. Therefore, efforts are needed to reduce noise which is implemented in LiDAR data processing software. In this study, the development of noise reduction technology that appears in the two-dimensional LiDAR data point cloud is carried out. The concept applied is the development of a systematic LiDAR data processing algorithm. The design of this algorithm contains visualization of object detection, storage of LiDAR data point cloud as detected object information, as well as noise reduction methods on the two-dimensional LiDAR data point cloud. This algorithm is realized in software form on Raspberry Pi 4 hardware, using the Python programming language. There are six algorithms used to reduce or eliminate noise, namely Algorithm 1, Algorithm 2, Algorithm 3, Algorithm 4, Algorithm 5, Algorithm 6. The experimental results show that the six algorithms created are able to display data visualization based on a 2-dimensional mapping system that is corrected for noise. The six algorithms succeeded in selecting noise up to 100%, although approximately 80% of the data that were considered correct could not be presented. Even if only 20% of the data is correct, the object structure is still recognizable
Copyrights © 2021