Azzam, Hamzah Nur (2025) Safe Overtaking pada Kendaraan Otonom Menggunakan NMPC dan Perencanaan Trajektori Bezier Real Time. Masters thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
6022222025-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (10MB) | Request a copy |
Abstract
Perkembangan teknologi di bidang keselamatan berkendara semakin pesat, khususnya dalam upaya mengatasi human error yang menjadi penyebab sekitar 75% kecelakaan lalu lintas, terutama saat melakukan manuver mendahului (overtaking). Penelitian ini mengusung metode Nonlinear Model Predictive Control (NMPC) untuk mengoptimalkan proses manuver overtaking yang membutuhkan kontrol adaptif dan komputasi yang cepat. Keunggulan NMPC terletak pada kemampuan prediksi (prediction horizon) yang memungkinkan sistem mengambil keputusan lebih cepat serta kompatibel dengan karakteristik sistem nonlinier. Dalam penelitian ini, fokus pengendalian diarahkan pada dua variabel utama, yaitu sudut kemudi (steering angle) dan kecepatan kendaraan. Trajektori manuver overtaking dirancang menggunakan kurva Bezier untuk menghasilkan jalur yang halus (smooth) dan dikembangkan agar mampu beradaptasi terhadap rintangan dinamis melalui mekanisme regenerasi lintasan secara real time. Simulasi dilakukan menggunakan ROS Noetic dan Gazebo, dengan sensor LiDAR Velodyne sebagai representasi perspektif pengemudi.
Hasil pengujian menunjukkan bahwa kombinasi NMPC dan kurva Bezier dinamis menghasilkan gerakan kendaraan yang lebih halus, aman, dan adaptif. Peningkatan nilai lookahead terbukti meningkatkan sensitivitas respons kendaraan, namun juga berdampak pada penurunan kestabilan ketika kendaraan bergerak pada lintasan yang berubah secara dinamis. Nilai prediction horizon terbaik diperoleh pada angka 15, setelah dilakukan pengujian pada rentang 5 hingga 25. Horizon yang terlalu besar menyebabkan ketidakstabilan pada lintasan melingkar, sedangkan horizon yang terlalu kecil mengakibatkan respons kendaraan menjadi lambat. Jarak aman terdekat antara sumbu pusat kendaraan terhadap rintangan terdekat tercatat sebesar 1,83 meter, dengan rata-rata jarak aman sebesar 1,98 meter. Dengan demikian, penelitian ini menjadi pendekatan yang realistis dalam membuktikan efektivitas metode NMPC dalam meningkatkan keselamatan manuver overtaking pada kendaraan otonom.
=======================================================================================================================================
The advancement of vehicle safety technology has grown rapidly, particularly in addressing human error, which accounts for approximately 75% of traffic accidents especially during overtaking maneuvers. This study proposes the Nonlinear Model Predictive Control (NMPC) method to optimize overtaking maneuvers, which require adaptive control and fast computation. NMPC offers advantages through its prediction horizon capability, enabling faster decision making and compatibility with nonlinear system characteristics. The control focus in this study is directed at two main varibles: steering angle and vehicle speed. The overtaking trajectory is designed using Bezier curves to generate smooth paths and is further developed to adapt to dynamic obstacles through real time trajectory regeneration. Simulations are conducted using ROS Noetic and Gazebo, with a Velodyne LiDAR sensor representing the driver's point of view. The test results show that the combination of NMPC and dynamic Bezier curves produces smoother, safer, and more adaptive vehicle movements. Increasing the lookahead value improves response sensitivity but reduces stability when navigating dynamically changing paths. The optimal prediction horizon is found to be 15, based on testing within the range of 5 to 25. A horizon that is too large leads to instability on curved paths, while a horizon that is too small results in delayed vehicle responses. The minimum safe distance between the vehicle's center axis and the nearest obstacle is recorded at 1.83 meters, with an average safe distance of 1.98 meters. Therefore, this study presents a realistic approach to demonstrating the effectiveness of NMPC in enhancing the safety of autonomous vehicle overtaking maneuvers.
Item Type: | Thesis (Masters) |
---|---|
Uncontrolled Keywords: | Nonlinear Model Predictive (NMPC), Mendahului, Mobil otonom, Kurva Bezier Nonlinear Model Predictive (NMPC), Overtaking, Autonomous Car, Bezier curve |
Subjects: | T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK1007 Electric power systems control |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis |
Depositing User: | Hamzah Nur Azzam |
Date Deposited: | 26 Jul 2025 13:14 |
Last Modified: | 26 Jul 2025 13:14 |
URI: | http://repository.its.ac.id/id/eprint/121690 |
Actions (login required)
![]() |
View Item |