Designing and Implementing Real-Time Deep Learning Object Detection in Unmanned Aerial Vehicles
Nurul Fahmi Arief Hakim, Alyani Ismail, Iwan Kustiawan, Deasy Rosanti, Taqiyuddin Yazid Zaidan · IIUM Engineering Journal · 2025
The rapid development of surveillance technology is readily apparent, particularly in monitoring vast and remote locations that present difficulties for human accessibility. Within the realm of contemporary surveillance methods, the utilization of Unmanned Aerial Vehicles (UAVs) has garnered a considerable amount of interest. The swift advancements in science and technology have led to the progressive incorporation of Artificial Intelligence (AI) technologies, particularly in tasks such as monitoring and reconnaissance. The results of this research contribute to the creation of a prototype for a UAV capable of conducting autonomous monitoring missions and integrating artificial intelligence technologies for real-time video processing. This study utilized an experimental methodology, and a Raspberry Pi was utilized for artificial intelligence processes integrated with the aircraft controller. During the decisive experiment, the unmanned aerial vehicle UAV could effectively travel to the designated area, reaching an accuracy of 78.6% in its AI processing. ABSTRAK: Perkembangan pesat teknologi pemantauan kini semakin ketara, terutamanya dalam pemantauan kawasan yang luas dan terpencil sukar diakses manusia. Dalam konteks kaedah pemantauan kontemporari, penggunaan Kenderaan Udara Tanpa Pemandu (UAV) telah menarik minat ramai. Kemajuan pesat sains dan teknologi telah menghasilkan gabungan teknologi Kecerdasan Buatan (AI) secara progresif, terutama dalam bidang pemantauan dan peninjauan. Penyelidikan ini memberi sumbangan kepada penciptaan prototaip UAV yang mampu menjalankan misi pemantauan sendiri, disamping gabungan teknologi kecerdasan buatan bagi memproses video masa nyata. Kajian ini menggunakan metodologi eksperimen dan Raspberry Pi yang disepadukan dengan pengawal pesawat bagi proses kecerdasan buatan. Semasa eksperimen penentuan, UAV berjaya bergerak dengan berkesan ke kawasan yang ditetapkan, mencapai ketepatan 78.6% dalam pemprosesan AI.