Skripsi
DETEKSI OBJEK DAN JALAN SECARA REAL TIME UNTUK KENDALI KEMUDI PADA AUTONOMOUS ELECTRIC VEHICLE BERBASIS DEEP LEARNING
Autonomous electric vehicles have a self-driving feature that can control the direction of the vehicle. This can be done through camera sensors as the input for steering control of the autonomous electric vehicle to detect roads and objects, and this method is called vision-based. However, previous research only focused on detecting either roads or objects, so in this study, the detection of roads and objects is combined as the input for steering control of the autonomous electric vehicle. The dataset used consists of 5 classes: roads, cars, motorcycles, people, and roadblocks, taken at Sriwijaya University's Indralaya campus. This study uses the YOLOv8 instance segmentation algorithm with the YOLOv8x-seg model trained for 100, 200, and 300 epochs. The best model was obtained at 200 epochs with the lowest segmentation loss of 0.53182. Then, testing was conducted through simulation, where the system was able to detect roads and objects accurately and measure object distances effectively. The implementation of the instance segmentation algorithm using YOLOv8 in real-time for identifying roads as the input for steering control of the autonomous electric vehicle was successfully performed, where the system was able to keep the autonomous electric vehicle on the road. Additionally, the system can be used to identify and measure the distance to objects with an average error of 2,33 meters. These distance measurements are used as input for the autonomous electric vehicle’s steering control to avoid objects.
Inventory Code | Barcode | Call Number | Location | Status |
---|---|---|---|---|
2307005519 | T122813 | T1228132023 | Central Library (Referens) | Available but not for loan - Not for Loan |
No other version available