Dynamic Path Planning Untuk Navigasi Robot Pelayan Restoran

Putra, Yuwana Perdana (2026) Dynamic Path Planning Untuk Navigasi Robot Pelayan Restoran. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022221029-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Restoran merupakan lingkungan kerja yang sempit, dinamis, dan padat interaksi, sehingga robot pelayan harus mampu bernavigasi secara aman sekaligus beradaptasi terhadap perubahan rintangan dengan cepat. Penelitian ini mengevaluasi enam algoritma global path planning (A*, Theta*, Hybrid A*, Dijkstra, D*, dan Voronoi) pada kerangka navigasi robot pelayan restoran holonomik RAISA berbasis peta occupancy grid 2D. Pengujian dilakukan pada skenario lingkungan restoran dengan rintangan dinamis yang berubah selama robot bergerak, serta menerapkan replanning berbasis peristiwa (event-based) dan berbasis waktu (time-based). Kinerja dari algoritma dianalisis menggunakan metrik distance jalur, latency perencanaan, time & travel, serta konsistensi lintasan pada variasi kondisi dan kecepatan.
Pada pengujian distance dan latency, skenario menggunakan jarak target 3 m, 5 m, dan 7 m dengan variasi posisi obstacle. Hasil uji latency menunjukkan Hybrid A* paling responsif dan stabil dengan kisaran sekitar 0.0028–0.0098 s, sedangkan Voronoi menjadi yang paling lambat (≈0.155–0.200 s) sehingga kurang ideal untuk replanning yang sering. Pada pengujian time & travel dengan variasi kecepatan 0.4–0.8 m/s, performa terbaik bergantung pada kecepatan: D* paling cepat pada 0.4 m/s (135.40 s), Theta* unggul pada 0.5 dan 0.7 m/s (109.28 s dan 82.62 s), sementara A* unggul pada 0.6 dan 0.8 m/s (89.86 s dan 76.33 s). Jarak total berada pada rentang 25.68–27.57 m dan tidak selalu sejalan dengan waktu tempuh, yang menunjukkan pengaruh karakter belokan dan kestabilan tracking. Untuk memperoleh solusi yang lebih konsisten pada skenario dinamis, penelitian ini juga menguji dua pendekatan dynamic path planning: kombinasi Hybrid A* dan Dijkstra dan kombinasi A* dan Hybrid A*. Kombinasi A* dan Hybrid A* memberikan keseimbangan terbaik dengan rata-rata jarak tempuh 25.842 m dan waktu tempuh total 75.11 s pada 0.8 m/s, serta lebih baik dibandingkan kombinasi Hybrid A* dan Dijkstra. Hal ini memberikan acuan praktis dalam pemilihan algoritma dan strategi replanning untuk robot pelayan restoran pada berbagai tingkat dinamika lingkungan dan kecepatan operasi.
====================================================================================================================================================================================
Restaurants are tight, dynamic, and interaction-dense environments, so a service robot must navigate safely while quickly adapting to changing obstacles. This study evaluates six global path-planning algorithms (A*, Theta*, Hybrid A*, Dijkstra, D*, and Voronoi) within a 2D occupancy-grid navigation framework for the holonomic RAISA restaurant service robot. Experiments are carried out in restaurant-like scenarios with dynamic obstacles that change as the robot moves, employing both event-based and time-based replanning. Algorithm performance is analyzed using path distance, planning latency, time-and-travel metrics, and trajectory consistency across different conditions and operating speeds. In the distance and latency tests, the scenarios use goal distances of 3 m, 5 m, and 7 m with varying obstacle placements. The latency results indicate that Hybrid A* is the most responsive and stable, ranging from approximately 0.0028 to 0.0098 s, while Voronoi is the slowest (≈0.155–0.200 s), making it less suitable for frequent replanning. In the time-and-travel tests at speeds of 0.4–0.8 m/s, the best-performing algorithm depends on speed: D* is fastest at 0.4 m/s (135.40 s), Theta* performs best at 0.5 and 0.7 m/s (109.28 s and 82.62 s), while A* performs best at 0.6 and 0.8 m/s (89.86 s and 76.33 s). The total travel distance ranges from 25.68 to 27.57 m and does not always align with travel time, indicating the influence of turning behavior and tracking stability. To obtain a more consistent solution in dynamic scenarios, this study also evaluates two dynamic path-planning approaches: a Hybrid A*–Dijkstra combination and an A*–Hybrid A* combination. The A*–Hybrid A* combination provides the best balance, achieving an average travel distance of 25.842 m and a total travel time of 75.11 s at 0.8 m/s, and it outperforms the Hybrid A*–Dijkstra combination. These results provide practical guidance for selecting algorithms and replanning strategies for restaurant service robots across different levels of environmental dynamics and operating speeds.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Dynamic Path Planning, Robot Pelayan, Service Robot, Occupancy Grid, Dijkstra, A*, Theta*, Hybrid A*, D*, Voronoi
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ211.415 Mobile robots
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Yuwana Perdana Putra
Date Deposited:	28 Jan 2026 07:25
Last Modified:	28 Jan 2026 07:25
URI:	http://repository.its.ac.id/id/eprint/130980

Actions (login required)

View Item