Syawalla, Andhika Ilham Syawalla (2026) Optimasi Penggunaan Energi pada Mobil Otonom Menggunakan Model Predictive Control. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5022221225-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5022221225-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (9MB) | Request a copy |
Abstract
Mobil otonom saat ini makin berkembang karena kemampuannya yang dapat mengambil keputusan sendiri tanpa intervensi manusia. Salah satu tantangan utama dalam pengembangan mobil otonom adalah kemampuan untuk mengikuti lintasan secara akurat dengan mempertimbangkan penggunaan energi. Pada penelitian tugas akhir ini, telah dirancang metode kontrol Model Predictive Control (MPC) yang dikombinasikan dengan PID Controller. Model Predictive Control digunakan untuk mengatur gerakan longitudinal dan lateral mobil otonom sekaligus meminimalkan konsumsi energi melalui optimasi daya. Pengujian sistem dilakukan menggunakn simulator CARLA dan ROS 2 pada lintasan perkotaan sepanjang 1 km dengan membandingkan variasi horizon prediksi 10, 20, dan 30 langkah. Hasil pengujian menun-jukan bahwa horizon prediksi 20 langkah merupakan konfigurasi optimum pada kondisi lalu lintas dinamis dengan menghasilkan galat pelacakan kecepatan terendah (RMSE 1,0629 m/s) serta daya rata-rata terendah (49,97 kW). Konfigurasi horizon pendek (10 langkah) meng-hasilkan gerakan kemudi paling stabil dengan penghematan energi mencapai 22.6 % pada kon-disi tanpa kendaraan lain, namun memicu pengereman longitudinal yang reaktif dan mendadak saat mengikuti kendaraan di depan. Sebaliknya, horizon panjang (30 langkah) menimbulkan osilasi kemudi frekuensi tinggi (steering chattering) akibat akumulasi galat linearisasi model prediksi, sehingga meningkatkan disipasi energi slip ban lateral secara signifikan.
===================================================================================================================================
Autonomous cars are currently expanding rapidly due to their ability to make decisions independently without human intervention. One of the main challenges in the development of autonomous cars is the ability to follow a trajectory accurately while considering energy consumption. In this final project, a Model Predictive Control (MPC) method combined with a PID controller is designed. Model Predictive Control is used to regulate both the longitudinal and lateral motions of the autonomous car while minimizing energy consumption through power optimization. System testing is conducted using the CARLA simulator and ROS 2 on a 1 km urban track, comparing prediction horizons of 10, 20, and 30 steps. The test results show that a prediction horizon of 20 steps is the optimal configuration in dynamic traffic conditions,
yielding the lowest speed tracking error (RMSE 1.0629 m/s) and the lowest average power (49.97 kW). A short prediction horizon of 10 steps produces the most stable steering control with energy savings up to 22.6% in obstacle-free conditions, but leads to reactive and abrupt longitudinal braking when following a lead vehicle. Conversely, a long prediction horizon of 30 steps triggers high-frequency steering oscillations (steering chattering) due to accumulated linearization errors in the prediction model, which significantly increases lateral tire slip energy dissipation
Actions (login required)
 |
View Item |