Implementasi Dead Reckoning Berbasis Deep Learning pada Mikrokontroler STM32 dengan Sensor IMU.

Azmi, Muhammad Rafie (2024) Implementasi Dead Reckoning Berbasis Deep Learning pada Mikrokontroler STM32 dengan Sensor IMU. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07211940000077_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2026. Download (11MB) | Request a copy

Abstract

Apakah kamu pernah mengalami saat sedang bernavigasi terdapat ketidaksesuaian posisi, seperti contohnya sekarang ada di posisi A namun pada GPS terbaca sedang di posisi B, perbedaan beberapa meter dapat berpengaruh pada dunia pemetaan geolokasi. Dalam bidang otonomi, beberapa robot sudah memiliki kemampuan untuk mendeteksi lokasi sebelumnya dan perpindahan ke lokasi saat ini. Dalam bidang olahraga, beberapa aplikasi merancang untuk penggunanya dapat memantau sudah berapa jauh berlari ataupun bersepeda. Saat berada dalam luar jangkauan GPS, maka navigator tidak bisa menjangkaunya, semua bisa dilakukan dengan pemrosesan perhitungan mati (Dead Reckoning) yang dijalankan menggunakan Machine Learning suatu program. Perhitungan mati tidak memerlukan akses GPS, hanya memerlukan sumbu-X, sumbu-Y, dan sumbu-Z dengan menggunakan catatan jalur yang telah dilalui, jarak yang ditempuh. Penggunaan unit pengukuran inertial (Inertial Measuring Unit) berupa akselerometer, magnetometer, dan giroskop. Nantinya keluaran data berupa posisi yang terbaca pada IMU akan diproses pada model jaringan saraf tiruan. Posisi sistem global (Global Position System) lebih unggul daripada sistem perhitungan mati dalam hal navigator, namun GPS memiliki banyak kekurangan seperti keterbatasan penetrasi sinyal dan keterlambatan penerimaan sinyal. Tetapi kedua sistem tersebut bisa dipadukan dan akan dibahas pada Tugas Akhir ini. Mencari model terbaik yang bisa dilakukan.
=================================================================================================================================
Have you ever experienced when navigating there is a position discrepancy, for example now it is in position A but on GPS it reads medium in position B, a difference of a few meters can affect the world of geolocation mapping. In the field of autonomy, some robots already have the ability to detect previous locations and move to the current location. In the field of sports, some applications design for users to monitor how far they have run or cycled. When outside GPS range, the navigator cannot reach it, all can be done with dead reckoning processing that is run using Machine Learning a program. Dead calculation does not require GPS access, only requires X-axis, Y-axis, and Z-axis using records of paths traveled, distances traveled. The use of inertial measuring units in the form of accelerometers, magnetometers, and gyroscopes. Later, the data output in the form of positions read on the IMU will be processed in the artificial neural network model. Global positioning systems are superior to dead calculation systems in terms of navigators, but GPS has many shortcomings such as signal penetration limitations and signal reception delays.But the two systems can be combined and will be discussed in this Final Project. Looking for the best model that can be done.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Perhitungan Mati, Unit Pengukuran Inertial, Jaringan Saraf Tiruan, Dead Reckoning, Inertial Measuring Unit, Artificial Neural Networks.
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7867.5 Noise T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7871.674 Detectors. Sensors
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Muhammad Rafie Azmi
Date Deposited:	09 Feb 2024 16:33
Last Modified:	09 Feb 2024 16:33
URI:	http://repository.its.ac.id/id/eprint/106551

Actions (login required)

View Item