Perancangan Algoritma Sensor Fusion Berbasis Filter Kalman Untuk Mendeteksi Kendaraan Menggunakan Kamera Stereo & Sensor Radar

Abdurrahman, Faishal (2021) Perancangan Algoritma Sensor Fusion Berbasis Filter Kalman Untuk Mendeteksi Kendaraan Menggunakan Kamera Stereo & Sensor Radar. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
02311640000155-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (1MB) | Request a copy

Abstract

Pada penelitian ini dilakukan perancangan algoritma sensor fusion berbasis filter kalman untuk mendeteksi suatu objek kendaraan dengan menggunakan kamera stereo dan sensor radar. Algoritma sensor fusion berbasis filter kalman bekerja dengan cara menginisialisasi, mengkonfirmasi, memprediksi, mengoreksi, dan menghapus track objek kendaraan yang bergerak. Pada perancangan ini digunakan kamera stereo dengan parameter intrinsik 2D yang terdiri dari focal length, principal point, image size dalam piksel yaitu sebesar ([1000 1000], [640 360], [720 1080]) dan sensor radar dengan parameter field of view yang terdiri dari azimuth dan elevasi sebesar 30º dan 5º. Pemodelan keadaan jalan dilakukan dengan parameter jalan datar dan lurus yang di dalamnya terdapat 4 lajur dengan 2 arah (panjang dan lebar jalan sebesar 100 m dan 22,15 m) dan parameter kendaraan dengan dimensi panjang, lebar, dan tinggi sebesar 4,63 m; 2,08 m; dan 1,416 m serta objek kendaraan yang dideteksi berjumlah tiga objek kendaraan. Keluaran algoritma sensor fusion adalah koordinat posisi objek dengan 10 variasi track deteksi dan didapatkan nilai rata-rata persentase error estimasi deteksi sebesar 0,36%. Hal tersebut sudah sesuai dengan nilai standar JCGM 100:2008 mengenai evaluation of measurement data yaitu dibawah 5%. ==================================================================================================================== In this study, a sensor fusion algorithm based on the Kalman filter was designed to detect a vehicle object using a stereo camera and radar sensor. The Kalman filter-based sensor fusion algorithm works by initializing, confirming, predicting, correcting, and deleting the track of moving vehicle objects. This design uses a stereo camera with 2D intrinsic parameters consisting of focal length, principal point, image size in pixels, namely ([1000 1000], [640 360], [720 1080]) and a radar sensor with field of view parameters consists of azimuth and elevation of 30º and 5º. Modeling of road conditions is carried out with flat and straight road parameters in which there are 4 lanes with 2 directions (road length and width of 100 m and 22.15 m) and vehicle parameters with dimensions of length, width and height of 4.63 m; 2.08 m; and 1,416 m and the detected vehicle objects amounted to three vehicle objects. The output of the sensor fusion algorithm is the coordinates of the object's position with 10 variations of detection tracks and the average percentage of detection error is 0.36%. This is in accordance with the JCGM standard value 100: 2008 regarding evaluation of measurement data, which is below 5%.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Algoritma sensor fusion, filter kalman, kamera stereo, sensor radar, sensor fusion algorithm based, kalman filter, stereo camera, radar sensor
Subjects: T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL152.8 Vehicles, Remotely piloted. Autonomous vehicles.
Divisions: Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User: Faishal Abdurrahman
Date Deposited: 06 Mar 2021 05:54
Last Modified: 06 Mar 2021 05:54
URI: https://repository.its.ac.id/id/eprint/83663

Actions (login required)

View Item View Item