Sistem Navigasi Multi-Robot Nonholonomic pada Lingkungan Kerja Dinamis Menggunakan Optimal Reciprocal Collision Avoidance

Athmar, Muhammad Mirza Fakhruddin (2026) Sistem Navigasi Multi-Robot Nonholonomic pada Lingkungan Kerja Dinamis Menggunakan Optimal Reciprocal Collision Avoidance. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022221125-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Perkembangan navigasi robot otonom nonholonomic di lingkungan kerja dinamis menuntut integrasi yang efisien antara perencana lintasan global yang optimal dan penghindaran rintangan lokal yang aman. Penelitian ini mengusulkan sistem navigasi pada differential drive mobile robot menggunakan algoritma IRRT* modifikasi (APF steering, adaptive step size, greedy pruning, dan penghalusan kurva Bézier G2CBS) sebagai perencana global. Kontroler pure pursuit dengan fitur backtrack corridor digunakan sebagai pelacak lintasan, serta sensor LiDAR dan algoritma ORCA modifikasi sebagai sistem penghindaran tabrakan lokal dinamis. Hasil pengujian menunjukkan bahwa varian APF-IRRT* Adaptive Greedy pada tingkat clearance 0,1 m berhasil memangkas waktu komputasi global secara signifikan sebesar 44,433% terhadap varian APF-IRRT* Adaptive, 78,07% terhadap APF-IRRT*, serta 3,464% terhadap IRRT*, dengan panjang jalur fisik terpendek mencapai 19,8262 m. Penerapan kurva Bézier G2CBS mampu mereduksi jumlah belokan tajam hingga 66,67% serta mempersingkat panjang lintasan hingga 0,6%. Pada pengujian pelacakan, lookahead distance 0,2 m menghasilkan akurasi tertinggi dengan nilai RMSE sebesar 0,003 m. Integrasi ORCA sukses mencapai kondisi zero collision di seluruh skenario dinamis dengan jarak aman minimal antar-robot sebesar 0,302 m hingga 0,408 m, serta berhasil memandu robot menghindari rintangan dinamis non-agen dan kembali ke jalur globalnya secara halus.
======================================================================================================================================
The development of nonholonomic autonomous robot navigation in dynamic work environments demands efficient integration between optimal global trajectory planners and safe local obstacle avoidance. This study proposes a navigation system on a differential drive mobile robot using a modified IRRT* algorithm (APF steering, adaptive step size, greedy pruning, and G2CBS Bézier curve smoothing) as a global planner. A pure pursuit controller with a backtrack corridor feature is used as a trajectory tracker, and a LiDAR sensor and a modified ORCA algorithm as a dynamic local collision avoidance system. The test results show that the APF-IRRT* Adaptive Greedy variant at a clearance level of 0.1 m successfully reduces global computation time by 44.433% compared to the APF-IRRT* Adaptive variant, 78.07% compared to APF-IRRT*, and 3.464% compared to IRRT*, with the shortest physical path length reaching 19.8262 m. The application of the G2CBS Bézier curve is able to reduce the number of sharp turns by up to 66,67% and shorten the path length by up to 0.6%. In tracking tests, a lookahead distance of 0.2 m produced the highest accuracy with an RMSE value of 0.003 m. ORCA integration successfully achieved zero collision conditions in all dynamic scenarios with a minimum safe distance between robots of 0.302 m to 0.408 m, and successfully guided the robot to avoid non-agent dynamic obstacles and return to its global path smoothly.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	IRRT* Modifikasi, Multi-Robot, ORCA, Penghindaran Tabrakan, Perencanaan Jalur, Collision Avoidance , IRRT* Modification, Multi-Robot, ORCA, Path Planning.
Subjects:	T Technology > T Technology (General) > T57.62 Simulation T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3070 Automatic control
Divisions:	Faculty of Electrical Technology > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Mirza Fakhruddin Athmar
Date Deposited:	24 Jul 2026 03:36
Last Modified:	24 Jul 2026 03:36
URI:	http://repository.its.ac.id/id/eprint/136767

Actions (login required)

View Item