Fitrayansyah, Alfian and Waru Djuriatno,, S.T., M.T. and Eka Maulana,, S.T., M.T., M.Eng. (2023) Guidance System Menggunakan Segmentasi Resnet-50 untuk Kendaraan Listrik Otonom. Sarjana thesis, Universitas Brawijaya.
Abstract
Kendaraan Listrik Otonom (KLO) merupakan kendaraan yang dilengkapi dengan kontrol otomatis sehingga dapat berjalan sendiri layaknya dikendalikan oleh manusia. Salah satu hal yang terpenting dalam pengembangan KLO adalah Artificial Intelligence (AI) karena dapat beroperasi secara adaptif. AI dalam KLO digunakan sebagai panduan dalam menjalankan kendali otomatis kendaraan atau guidance system. Banyak sensor yang digunakan dalam guidance system, seperti LiDAR, Radar, Kamera Termal, serta Encoder. Permasalahan yang sering terjadi pada penggunaan kamera biasa adalah apabila terjadi kondisi-kondisi tertentu seperti adanya kabut, malam hari, dan lain-lain, maka akan mengganggu hasil yang didapat. Permasalahan ini dapat diselesaikan dengan penggunaan kamera termal karena kamera termal dapat berfungsi dengan baik pada berbagai kondisi lingkungan dengan visibilitas rendah dan kontras yang tinggi. Contoh visibilitas rendah yaitu paparan sinar matahari langsung, paparan lampu depan, kabut, dan asap. Sedangkan contoh kontras yang tinggi adalah kondisi malam hari, bayangan, matahari terbenam, dan matahari terbit. Kamera termal dapat digunakan untuk deteksi objek dan segmentasi semantik. Gambar yang diperoleh kamera termal pada KLO dapat dimanfaatkan untuk mensegmentasikan letak mobil, jalan, pohon, bangunan, dll pada tiap piksel. Hasil segmentasi jalan lintasan KLO selanjutnya digunakan untuk menentukan arah steering. Prosesnya berlangsung dengan melalui pemanfaatan library OpenCV. Proses ini digunakan untuk mengubah segmentasi jalan menjadi warna putih, sedangkan segmentasi yang lainnya akan berubah warna menjadi hitam. Karena warna putih bernilai 255, dan warna hitam bernilai 0, maka dapat dikalkulasikan dan diperoleh perbandingan nilai dari setiap piksel gambar segmentasi jalan antara badan jalan bagian kanan dan kiri gambar segmentasi jalan. Sehingga penelitian ini akan membahas tentang rancang bangun guidance system penentuan arah steering dengan memanfaatkan hasil segmentasi jalan berdasarkan informasi yang diperoleh dari kamera termal. Setelah dilakukan percobaan, guidance system melalui penentuan arah steering berdasarkan hasil segmentasi jalan menggunakan hasil deploy model FCN- Resnet-50 berhasil dilakukan melalui beberapa tahapan. Hasil rekomendasi steering berupa perintah kanan, kiri, dan tengah, serta presentase rekomendasi arah yang ditampilkan melalui LED dan LCD. Kesalahan pemberian rekomendasi steering diakibatkan model FCN-Resnet-50 yang digunakan masih banyak terjadi kesalahan segmentasi.
English Abstract
Autonomous Electric Vehicles (AEV) are vehicles that are equipped with automatic controls so that they can run independently as if controlled by humans. One of the most important things in the development of AEV is Artificial Intelligence (AI) because it can operate adaptively. AI in AEV is used as a guide in carrying out automatic vehicle control or guidance systems. Many sensors are used in the guidance system, such as LiDAR, Radar, Thermal Camera, and Encoder. The problem that often occurs with the use of ordinary cameras is that when certain conditions occur, such as fog, night, etc., it will interfere with the results obtained. This problem can be solved by using a thermal camera because thermal cameras can function well in various environmental conditions with low visibility and high contrast. Examples of low visibility are exposure to direct sunlight, exposure to headlights, fog, and smoke. Meanwhile, examples of high contrast are night conditions, shadows, sunsets, and sunrises. Thermal cameras can be used for object detection and semantic segmentation. The image obtained by the thermal camera on the AEV can be used to segment the location of cars, roads, trees, buildings, etc. at each pixel. The results of the AEV route segmentation are then used to determine the steering direction. The process takes place by utilizing the OpenCV library. This process is used to change the road segmentation to white, while the other segmentation will change color to black. Because the white color has a value of 255, and the black color has a value of 0, it can be calculated and obtained as a comparison of the values of each pixel of the road segmentation image between the right and left sides of the road segmentation image. So this research will discuss the design and construction of the guidance system for determining the direction of the steering by utilizing the results of road segmentation based on information obtained from the thermal camera. After the experiments, the guidance system through determining the direction of the steering based on the results of road segmentation using the results of deploying the FCN-Resnet-50 model was successfully carried out through several stages. The results of the steering recommendations are in the form of commands for right, left, and center, as well as the percentage of direction recommendations displayed via the LED and LCD. The error in giving steering recommendations is caused by the FCN-Resnet-50 model used which still has a lot of segmentation faults.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | 0523070007 |
| Uncontrolled Keywords: | Kendaraan listrik otonom, Guidance system, FCN-Resnet-50, Steering, Segmentasi Semantik .- Autonomous electric vehicle, Guidance system, FCN-Resnet-50, Steering, Semantic Segmentation |
| Subjects: | 600 Technology (Applied sciences) > 621 Applied physics > 621.3 Electrical, magnetic, optical, communications, computer engineering; electronics, lighting > 621.38 Electronics, communications engineering > 621.381 Electronics |
| Divisions: | Fakultas Teknik > Teknik Elektro |
| Depositing User: | Endang Susworini |
| Date Deposited: | 05 Jun 2023 07:01 |
| Last Modified: | 05 Jun 2023 07:01 |
| URI: | http://repository.ub.ac.id/id/eprint/200737 |
![[thumbnail of DALAM MASA EMBARGO]](http://repository.ub.ac.id/style/images/fileicons/text.png) |
Text (DALAM MASA EMBARGO)
Alfian Fitrayansyah.pdf Restricted to Registered users only until 31 December 2025. Download (3MB) |
Actions (login required)
 |
View Item |