Muhammad Bagas Hidayatullah, - (2025) DESAIN DAN IMPLEMENTASI BEHAVIOR CLONING SEBAGAI LANE KEEPING ASSIST KENDARAAN OTONOM SATU PENUMPANG. S1 thesis, Universitas Pendidikan Indonesia.
Abstract
Kendaraan otonom semakin diminati, dengan pasar yang terus berkembang serta dukungan regulasi dari pemerintah global maupun Indonesia yang mendorong kemajuan teknologi ini. Namun, biaya pengembangan kendaraan otonom tetap menjadi tantangan utama. Metode imitation learning mengurangi kebutuhan pemrograman manual yang kompleks dan pemodelan yang rumit, sehingga membantu menekan biaya pengembangan secara keseluruhan. Penelitian ini mengimplementasikan imitation learning untuk lane keeping assist pada kendaraan otonom berpenumpang tunggal dalam lingkungan bebas hambatan menggunakan simulasi ROS dan kendaraan O-SEATER dengan sistem penggerak diferensial. Model dikembangkan dengan kombinasi Convolutional Neural Networks (CNN) dan Fully Connected Networks (FCN) untuk mengekstraksi fitur data masukan berupa gambar dari kamera depan. Model ini dirancang untuk memprediksi angular velocity kendaraan, yang secara langsung mengontrol arah pergerakan berdasarkan kondisi lingkungan. Pendekatan Behavior cloning digunakan untuk memahami dan mereplikasi perilaku pengemudi berdasarkan data historis yang diperoleh dari simulasi dan pengujian dunia nyata. Hasil eksperimen menunjukkan bahwa model ini mampu mencapai performa yang cukup baik, dengan pengujian Mean Squared Error (MSE) yang mendekati nol. Penelitian ini membuktikan bahwa pendekatan Behavior cloning berbasis CNN dan FCN dapat diterapkan untuk lane keeping assist pada kendaraan otonom, dengan hasil yang baik dalam simulasi maupun pengujian pada kondisi pagi, siang, dan sore. ---------- Autonomous vehicles are becoming increasingly popular, with a growing market and regulatory support from both global and Indonesian governments driving advancements in this technology. However, the high development costs of autonomous vehicles remain a significant challenge. The imitation learning method reduces the need for complex manual programming and modeling, thereby lowering overall development costs. This study implements imitation learning for lane keeping assist in a single-passenger autonomous vehicle within an obstaclefree environment using ROS simulation and the O-SEATER vehicle with a differential drive steering system. The model is developed using a combination of Convolutional Neural Networks (CNN) and Fully Connected Networks (FCN) to extract features from input data in the form of front camera images. The model is designed to predict the vehicle's angular velocity, directly controlling its direction of movement based on environmental conditions. The Behavior cloning approach is employed to understand and replicate driver behavior based on historical data collected from simulations and real-world testing. Experimental results show that the model achieves satisfactory performance, with a Mean Squared Error (MSE) close to zero. This study demonstrates that the Behavior cloning approach based on CNN and FCN can be effectively applied for lane keeping assist in autonomous vehicles, yielding favorable results in simulations and testing under morning, afternoon, and evening conditions.
![[img]](http://repository.upi.edu/style/images/fileicons/text.png) |
Text
S_TEKKOM_2104096_Title.pdf Download (936kB) |
|
Text
S_TEKKOM_2104096_Chapter1.pdf Download (143kB) |
|
Text
S_TEKKOM_2104096_Chapter2.pdf Restricted to Staf Perpustakaan Download (401kB) | Request a copy |
|
Text
S_TEKKOM_2104096_Chapter3.pdf Download (733kB) |
|
Text
S_TEKKOM_2104096_Chapter4.pdf Restricted to Staf Perpustakaan Download (6MB) | Request a copy |
|
Text
S_TEKKOM_2104096_Chapter5.pdf Download (41kB) |
|
Text
S_TEKKOM_2104096_Appendix.pdf Restricted to Staf Perpustakaan Download (587kB) | Request a copy |
| Item Type: | Thesis (S1) |
|---|---|
| Uncontrolled Keywords: | Imitation Learning; Behavior cloning; lane keeping assist; Kendaraan Otonom. |
| Subjects: | T Technology > T Technology (General) T Technology > TA Engineering (General). Civil engineering (General) T Technology > TL Motor vehicles. Aeronautics. Astronautics |
| Divisions: | UPI Kampus cibiru > S1 Teknik Komputer |
| Depositing User: | Muhammad Bagas Hidayatullah |
| Date Deposited: | 05 Mar 2025 03:22 |
| Last Modified: | 05 Mar 2025 03:22 |
| URI: | http://repository.upi.edu/id/eprint/130148 |
Actions (login required)
 |
View Item |