Navigasi Robot Beroda Otonom Berbasis Logika Fuzzy Dengan Fusi Sensor Lidar Dan Kamera Termal

Alfalakhi, Priagung Ramadhan (2026) Navigasi Robot Beroda Otonom Berbasis Logika Fuzzy Dengan Fusi Sensor Lidar Dan Kamera Termal. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022221161-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (13MB) | Request a copy

Abstract

Navigasi yang aman merupakan salah satu aspek penting pada robot bergerak otonom, terutama ketika robot beroperasi pada lingkungan yang mengandung obstacle statis maupun dinamis. Penggunaan LiDAR 2D sebagai sensor utama navigasi memiliki keterbatasan dalam mendeteksi obstacle berprofil rendah yang berada di luar bidang pemindaian sehingga menimbulkan area blind-spot dan meningkatkan risiko tabrakan. Penelitian ini merancang sistem navigasi robot beroda otonom berbasis logika fuzzy Mamdani dengan fusi sensor LiDAR 2D dan kamera termal untuk meningkatkan kualitas persepsi lingkungan dan keselamatan navigasi. Sistem diimplementasikan pada robot beroda holonomik dengan footprint 0,54 m × 0,64 m menggunakan ROS Noetic dan disimulasikan pada simulator Gazebo. Informasi geometris dari LiDAR dan informasi termal dari kamera diolah menjadi traversability LiDAR dan traversability termal yang kemudian digabungkan melalui pendekatan decision-level fusion untuk menghasilkan effective traversability. Nilai tersebut digunakan sebagai masukan logika fuzzy Mamdani untuk menyesuaikan parameter navigasi Artificial Potential Field (APF) secara adaptif berdasarkan kondisi lingkungan yang diamati robot. Sistem dievaluasi pada delapan skenario pengujian dengan total 80 percobaan yang mencakup lingkungan terbuka, obstacle statis, obstacle berprofil rendah pada area blind-spot LiDAR, serta obstacle dinamis. Pada skenario yang mengandung obstacle berprofil rendah dan obstacle dinamis, metode yang diusulkan meningkatkan Success Rate dan Collision-Free Rate dari 0–65% pada konfigurasi baseline menjadi 100%. Nilai minimum clearance juga meningkat lebih dari lima kali lipat, yaitu dari 0,035 m menjadi 0,18 m, serta berubah dari kondisi kontak fisik (−0,16 m) menjadi jarak aman positif (0,32 m). Uji Mann–Whitney U menunjukkan peningkatan tersebut signifikan secara statistik dengan ukuran efek besar (r = 1,00). Secara kualitatif, robot melakukan manuver penghindaran lebih awal dan menghasilkan pergerakan yang lebih halus. Meskipun metode yang diusulkan menghasilkan peningkatan waktu tempuh dan panjang lintasan, peningkatan tersebut masih berada dalam batas yang wajar dibandingkan dengan margin keselamatan yang diperoleh. Berdasarkan hasil tersebut, dapat disimpulkan bahwa integrasi kamera termal, fusi traversability, dan logika fuzzy Mamdani pada sistem navigasi berbasis APF mampu mengatasi keterbatasan blind-spot LiDAR sekaligus meningkatkan keselamatan navigasi robot bergerak otonom dibandingkan pendekatan yang hanya mengandalkan LiDAR.
===============================================================================================================================
Safe navigation is a critical requirement for autonomous mobile robots, particularly when operating in environments containing static and dynamic obstacles. Although 2D LiDAR is widely used as a primary navigation sensor due to its accurate distance measurement capability, it suffers from limitations in detecting low-profile obstacles located outside its scanning plane, resulting in blind-spot regions that increase the risk of collision. This research designs an autonomous wheeled robot navigation system based on Mamdani fuzzy logic and 2D LiDAR-thermal camera sensor fusion to improve environmental perception and navigation safety. The system is implemented on a holonomic wheeled robot with a footprint of 0.54 m × 0.64 m using ROS Noetic and simulated in the Gazebo simulator. Geometric information from the LiDAR and thermal information from the camera are processed into LiDAR traversability and thermal traversability, which are subsequently combined through a decision-level fusion approach to generate an effective traversability value. This value is used as the input of a Mamdani fuzzy logic controller to adaptively adjust the parameters of the Artificial Potential Field (APF) navigation method according to the observed environmental conditions. The system is evaluated in the Gazebo simulator through eight testing scenarios with a total of 80 trials, involving open environments, static obstacles, low-profile obstacles located within LiDAR blind-spot regions, and dynamic obstacles. In scenarios involving low-profile and dynamic obstacles, the proposed method improves the Success Rate and Collision-Free Rate from 0–65% in the baseline configuration to 100%. The minimum clearance also increases by more than fivefold, from 0.035 m to 0.18 m, and shifts from a physical-contact condition (−0.16 m) to a positive safety distance (0.32 m). A Mann–Whitney U test confirms that these improvements are statistically significant with a large effect size (r = 1.00). Qualitatively, the robot performs earlier avoidance maneuvers and produces smoother movement. Although the proposed method results in increased traversal time and path length, these increases remain within reasonable bounds compared with the safety margin gained. These results indicate that the integration of a thermal camera, traversability-based sensor fusion, and Mamdani fuzzy logic within an APF-based navigation framework effectively overcomes the LiDAR blind-spot limitation while enhancing the navigation safety of autonomous mobile robots compared with a navigation system relying solely on LiDAR.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	robot bergerak otonom; LiDAR 2D; kamera termal; traversability; fusi sensor; logika fuzzy Mamdani; Artificial Potential Field; autonomous mobile robot; 2D LiDAR; thermal camera; traversability; sensor fusion; Mamdani fuzzy logic; Artificial Potential Field
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ211.415 Mobile robots
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Priagung Ramadhan Alfalakhi
Date Deposited:	20 Jul 2026 03:00
Last Modified:	20 Jul 2026 03:00
URI:	http://repository.its.ac.id/id/eprint/135536

Actions (login required)

View Item