Tracking Control Of Autonomous Car With Attention To Obstacle Using Model Predictive Control

Alya, Yasmina (2022) Tracking Control Of Autonomous Car With Attention To Obstacle Using Model Predictive Control. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07111840000107-Undergraduate_Thesis.pdf Download (4MB)

Abstract

Pada penelitian terdahulu mengenai MPC untuk path tracking dan obsctacle avoidance, menunjukan MPC menghasilkan gerakan berosilasi pada path tracking dan terkadang MPC tidak mampu menghindari halangan dengan manuver yang efektif. Penelitian dilakukan dengan memvariasikkan bobot cost function dan hasil terbaik diperoleh ketika bobot eror lebih besar dibandingkan dengan bobot input. Penelitian tersebut dibatasi dengan kecepatan mobil konstan. Maka pada penelitian ini penulis bertujuan untuk menggunakan MPC untuk trajectory tracking dan obstacle avoidance. Pada penelitian ini, digunakan referensi yang berubah terhadap waktu (trajectory tracking). Sehingga MPC juga akan mengatur kecepatan mobil. Kombinasi parameter MPC divariasikan untuk mencari desain MPC dengan performa baik. Penelitian ini menggunakan Linear Time Variant MPC (LTV MPC). Pada sistem obstacle avoidance, deteksi halangan dilakukan dengan mengukur jarak antara mobil dengan halangan. Selama halangan terdeteksi, dilakukan perhitungan constraint posisi Y baru. Pengujian trajectory tracking dilakukan dengan 3 referensi track sinusoidal dan 1 track pergantian lajur. Pengujian obstacle avoidance menggunakan 1 dan 2 halangan. Evaluasi dilakukan berdasarkan nilai RMSE posisi, cost function, dan jarak terdekat dengan halangan saat manuver penghindaran. Hasil menunjukan MPC berhasil menjalankan fungsi trajectory tracking dengan keterlambatan rata rata 0.4 s, sehingga lebih cocok digunakan untuk kecepatan rendah. MPC tidak selalu berhasil melakukan obstacle avoidance karena terdapat variasi yang tidak memenuhi zona aman. Dapat diimplikasikan bahwa perumusan halangan sebagai constraint posisi Y kurang cocok digunakan untuk menjalankan fungsi obstacle avoidance pada MPC. Karena pada obstacle avoidance dihasilkan Gerakan osilasi yang belum cocok untuk implementasi dunia nyata.
====================================================================================================================================
Previous research about MPC in for autonomous car shows that MPC generates an oscillating movement when doing path tracking and can’t avoid obstacle effectively. The research is done by varying cost function weights where the best performance is produced when the error weight is greater than the input weight. Limitations of research includes the usage of constant velocity. Therefore, in this research the writer aims to use MPC for trajectory tracking and obstacle avoidance. So MPC will also controls the car’s velocity. Time varying reference is used to represent trajectory tracking problem. Combinations of MPC parameters are varied to search for the best performing design. Research is done using Linear Time Variant MPC (LTV MPC). In obstacle avoidance system, obstacle detection is done by calculating the distance between the car and obstacle. While an obstacle is detected, the system recalculates a new lateral constraint to ensure the car move within a predefined safe zone. Trajectory tracking tests are done using 3 sinusoidal tracks and a lane change maneuver track. Obstacle avoidance tests are done using 1 and 2 obstacles. Evaluation is based on RMSE of position, cost function, and the closest distance between car and obstacle during avoidance maneuver. Results shows MPC can be used for trajectory tracking with average delay of 0.4 s. So it is better for a lower velocity. MPC can’t always avoid obstacle, shown by one of the variation that can’t fulfill the safe zone. It can be implicated that defining obstacle as lateral constraint is not suitable for obstacle avoidance function. The oscillating car movement shows that the system is not yet suitable for real life implementation.

Item Type:	Thesis (Other)
Additional Information:	RSE 629.893 2 Aly t-1 2022
Uncontrolled Keywords:	Mobil otonom, Model Predictive Control, Path Tracking, Obstacle Avoidance, Autonomous car, Model Predictive Control, Path Tracking, Obstacle Avoidance
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ217 Adaptive control systems T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL152.8 Vehicles, Remotely piloted. Autonomous vehicles.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	- Davi Wah
Date Deposited:	12 Sep 2024 05:44
Last Modified:	12 Sep 2024 05:44
URI:	http://repository.its.ac.id/id/eprint/115543

Actions (login required)

View Item