Sabiq, Muhammad Faqih Tajus (2022) Fusi Radar dan Kamera Menggunakan Centerfusion untuk Deteksi Objek pada Kendaraan Otonom. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 07111840000043-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
07111840000043-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (2MB) |
Abstract
Kendaraan otonom biasanya dilengkapi dengan berbagai jenis sensor untuk memanfaatkan karakteristik pelengkapnya. Beberapa modalitas sensor yang digunakan di dalam kendaraan otonom dapat meningkatkan robustness dan accuracy . Namun, penggunaan tersebut juga memperkenalkan tantangan baru dalam merancang sistem persepsi. Sistem persepsi dalam kendaraan otonom bertanggung jawab untuk mendeteksi dan men-tracking objek sekeliling. Untuk memproses arus data yang dihasilkan oleh sensor diperlukan algoritma. CenterFusion adalah algoritma yang menggunakan center point detection network untuk mendeteksi objek dengan mengidentifikasi titik tengah pada gambar. Selanjutnya titik tengah pada gambar diasosiasikan dengan deteksi radar dengan menggunakan metode berbasis frustum. Deteksi radar yang sudah diasosiasikan kemudian digunakan untuk menghasilkan feature-map berbasis radar untuk melengkapi image feature dan meregresi properti dari objek seperti depth, rotation, dan velocity. Dibandingkan CenterNet, algoritma yang hanya menggunakan kamera, penggunaan radar dan dalam algoritma CenterFusion dapat menaikkan nilai AP dari class objek secara keseluruhan.
==============================================================================================================================
Autonomous vehicles are usually equipped with various types of sensors to utilize their complementary characteristics. Some sensor modalities used in autonomous vehicles can increase robustness and accuracy. However, such use also introduces new challenges in designing perception systems. The perception system in autonomous vehicles is responsible for detecting and tracking surrounding objects. To process the data stream generated by sensors, an algorithm is required. CenterFusion is an algorithm that uses a center point detection network to detect objects by identifying the center point in an image. Furthermore, the center point in the image is associated with radar detection using a frustum-based method. The associated radar detection is then used to generate a radar-based feature-map to complement image features and regress object properties such as depth, rotation, and velocity. Compared to CenterNet, an algorithm that only uses a camera, the use of radar within the CenterFusion algorithm can increase the AP value of object classes overall.
| Item Type: | Thesis (Other) |
|---|---|
| Additional Information: | RSE 621.367 Sab f-1 2022 |
| Uncontrolled Keywords: | Sensor Fusion, Object Detection, Autonomous Vehicle. |
| Subjects: | T Technology > TK Electrical engineering. Electronics Nuclear engineering |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis |
| Depositing User: | Mr. Marsudiyana - |
| Date Deposited: | 12 Jun 2026 02:57 |
| Last Modified: | 12 Jun 2026 02:57 |
| URI: | http://repository.its.ac.id/id/eprint/133758 |
Actions (login required)
 |
View Item |