Development of Lidar-Based Object Detection Software

Abstract

The car is a means of transportation whose needs are getting higher. This situation is not only happening abroad but also in Indonesia. However, the car's existence is currently complained about because of the pollution it produces and the level of comfort. Expectations in the future seem to be more directed at electric cars with deficient pollution levels and convenience in their use, such as autonomous electric cars. Therefore, we must master the technology of autonomous electric vehicles to enter the era of Mobility in Society 5.0. One form of related technology is a LiDAR-based object detection software system. Sometimes, the software accompanying a tool cannot provide various facilities according to the field's application needs. This condition is due to the manufacturer's limitations, likewise with 2D LiDAR products widely marketed, for example, YDLiDAR. For object detection applications, the provided software has limitations in data storage, flexibility in data presentation, and the ability to reduce noise when the LiDAR is operated under certain conditions. This study developed LiDAR-based object detection software to overcome the shortcomings mentioned above by adding the functions discussed above and applying them to object detection and distance recognition. In general, this system combines a software system developed on a laptop with a hardware system consisting of YDLiDAR G4 and a serial data interface. This software system was also created using the python programming language. The measurement results show that the performance of the software developed has good visual performance. This software can store detection data for a specified duration and suppress noise, which is quite good. The noise reduction capability of this software system can reduce errors up to 19.2%.