Wicaksono, M. Januar Eko (2026) Pengembangan Sistem Drone Autonomous Delivery Dengan Web-Based Waypoint Planning, GPS-Based Precision Landing, Dan Customer-Driven Package Verification. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5027221006-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5027221006-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (1MB) | Request a copy |
Abstract
Pertumbuhan e-commerce dan tantangan last-mile delivery kemacetan di wilayah urban serta keterbatasan akses di daerah terpencil mendorong kebutuhan akan solusi pengiriman yang lebih cepat dan efisien. Drone autonomous delivery menjadi alternatif yang menjanjikan, namun implementasinya masih terkendala akurasi pendaratan, verifikasi paket, dan konfirmasi penerimaan pada konteks operasi otonom. Penelitian ini mengembangkan sistem drone autonomous delivery terintegrasi end-to-end yang menyatukan tiga pilar utama, yaitu web-based waypoint planning, GPS-based precision landing, dan customer-driven package verification. Sistem dibangun pada quadcopter F450 dengan flight controller Pixhawk 2.4.8 (ArduCopter) dan Raspberry Pi 4 sebagai companion computer yang berkomunikasi melalui protokol MAVLink. Antarmuka operator dan penerima dikembangkan sebagai aplikasi web berbasis React dengan peta interaktif Leaflet/OpenStreetMap, didukung backend Node.js/Express, komunikasi real-time Socket.IO, basis data Supabase, serta live camera streaming HLS berbasis FFmpeg yang diekspos melalui Cloudflare Tunnel. Hasil pengujian menunjukkan navigasi waypoint berbasis web mencapai rata-rata error 1,789 meter (RMSE 1,851 meter) dengan tingkat keberhasilan 100% pada sepuluh pengujian. Precision landing berbasis GPS murni menghasilkan rata-rata deviasi 1,323 meter (memenuhi target ≤3 meter) dengan RMSE 1,348 meter dan akurasi terbaik 0,94 meter pada kondisi HDOP rendah; seluruh pengujian dilaksanakan pada kondisi angin tenang (2 hingga 3 m/s) sehingga akurasi terbukti dipengaruhi secara signifikan oleh kualitas sinyal GPS (HDOP). Mekanisme konfirmasi penerima beserta pelepasan paket melalui Servo berhasil pada seluruh pengujian (100%), sedangkan pengujian end-to-end mencapai keberhasilan 80%. Secara keseluruhan, seluruh sepuluh komponen sistem (100%) berfungsi sesuai rancangan, membuktikan kelayakan sistem sebagai proof-of-concept yang menempatkan penerima sebagai bagian aktif dalam alur pengiriman.
===================================================================================================================================
The growth of e-commerce and the challenges of last-mile delivery urban traffic congestion and limited access in remote areas drive the need for faster and more efficient delivery solutions. Autonomous drone delivery is a promising alternative, yet its implementation still faces challenges in landing accuracy, package verification, and receipt confirmation within autonomous operations. This research develops an end-to-end integrated autonomous drone delivery system that unifies three main pillars, that are web-based waypoint planning, GPS-based precision landing, and customer-driven package verification. The system is built on an F450 quadcopter with a Pixhawk 2.4.8 flight controller (ArduCopter) and a Raspberry Pi 4 companion computer communicating via the MAVLink protocol. The operator and recipient interfaces are developed as a React web application with an interactive Leaflet/OpenStreetMap map, supported by a Node.js/Express backend, real-time Socket.IO communication, a Supabase database, and FFmpeg-based HLS live camera streaming exposed through Cloudflare Tunnel. Testing shows that web-based waypoint navigation achieved a mean error of 1.789 meters (RMSE of 1.851 meters) with a 100% success rate over ten trials. GPS-only precision landing produced a mean deviation of 1.323 meters (meeting the ≤3-meter target) with an RMSE of 1.348 meters and a best accuracy of 0.94 meters under low-HDOP conditions; all trials were conducted under calm wind conditions ( 2 hingga 3 m/s), so accuracy was significantly affected by GPS signal quality (HDOP) alone. The recipient confirmation mechanism and Servo-based package release succeeded in all trials (100%), while end-to-end testing reached a 80% success rate. Overall, all ten system components (100%) functioned as designed, demonstrating the feasibility of the system as a proof-of-concept that positions the recipient as an active part of the delivery flow.
Actions (login required)
 |
View Item |