Hand Pose Recognition Dengan Deep Learning Berbasis Komputasi Edge Untuk Kendali Drone DJI Tello

Syahputra, Ardy Pratama (2025) Hand Pose Recognition Dengan Deep Learning Berbasis Komputasi Edge Untuk Kendali Drone DJI Tello. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5024211035-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (23MB) | Request a copy

Abstract

Penelitian ini mengembangkan sistem kendali drone DJI Tello menggunakan pose tangan berbasis metode Deep Learning yang diimplementasikan pada komputasi edge menggunakan Raspberry Pi 5. Sistem ini memungkinkan pengguna untuk mengendalikan drone secara langsung tanpa remote kontrol konvensional, menawarkan alternatif kontrol yang praktis dan intuitif. Model MLP terbukti paling efisien dengan waktu inferensi tercepat sebesar 0,13 ms serta frame per second (fps) tertinggi, berkisar antara 26,99 hingga 29,68 fps dibandingkan dengan model CNN 1D dan CNN 2D. Model MLP dan CNN 1D juga menunjukkan performa akurasi yang unggul pada berbagai jarak dan kondisi pencahayaan rendah dibandingkan dengan CNN 2D. Secara keseluruhan, sistem ini mampu beroperasi secara optimal di kondisi pencahayaan cukup dan tetap efektif pada berbagai pengguna dengan tingkat keberhasilan di atas 90%.
==================================================================================================================================
This research develops a DJI Tello drone control system utilizing hand pose recognition based on Deep Learning, implemented on edge computing using Raspberry Pi 5. The system allows users to control the drone directly without a conventional remote control, offering a practical and intuitive alternative control method. The MLP model demonstrated the highest efficiency with the fastest inference time of 0.13 ms and the highest frames per second (fps), ranging from 26.99 to 29.68 fps, compared to CNN 1D and CNN 2D models. Additionally, MLP and CNN 1D models showed superior accuracy performance across various distances and lowlight conditions compared to the CNN 2D model. Overall, the system operates optimally under adequate lighting conditions and remains effective across different users, achieving success rates above 90%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Pose tangan, Deep Learning, MLP, CNN 1D, CNN 2D, Raspberry Pi 5, Kendali drone, Komputasi edge, Hand Pose, Drone Control, Edge Computing
Subjects:	T Technology > T Technology (General) > T11 Technical writing. Scientific Writing T Technology > T Technology (General) > T385 Visualization--Technique T Technology > T Technology (General) > T57.5 Data Processing T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Ardy Pratama Syahputra
Date Deposited:	16 Jun 2025 07:09
Last Modified:	16 Jun 2025 07:09
URI:	http://repository.its.ac.id/id/eprint/119163

Actions (login required)

View Item