PERANCANGAN PURWARUPA ALAT DETEKSI KERUSAKAN JALAN BERBASIS MACHINE LEARNING
Abstract
The lengthy process of road damage detection is one of the cause of delays of road repair
processes in Indonesia. Manual detection method takes a lot of time and is highly prone to errors.
Although several technologies have been developed to solve this problem, the high cost and
complexity of use often become significant barriers. This research aims to design a prototype of
a low-cost road damage detection tool that can help accelerate the process of road damage
detection and inspection in Indonesia. The system uses remote sensing principles with YOLOv8s
Object Detection technology to detect four classes of road damage. The prototype was designed
using alternative electronic components powered by Raspberry Pi 5 and Hilo-8L accelerator. and
Hailo-8L accelerator. The use of this alternative device is capable of running Object Detection
model as well as other computational tasks stably with an average of ?27 FPS. The Ublox GPS
Module Neo-7M GPS Module used in the system has a deviation of ±3.5 meters. The designed
object detection model has an overall accuracy of 92,96%. Testing results of the prototype in
stationary conditions showed an optimal angle of 20-30 degrees with detection distances of 1, 3,
and 5 meters. In moving conditions, optimal system performance is at a speed of 20-30 km/h with
effective light intensity in the morning and afternoon.
Downloads
This work is licensed under a Creative Commons Attribution 4.0 International License.
