Evaluasi Robustness Sistem Localization Mobile Robot Berbasis EKF dan AMCL Terhadap Degradasi Sensor dan Gangguan Lingkungan

Garninto, Affief Adli Taqy (0022) Evaluasi Robustness Sistem Localization Mobile Robot Berbasis EKF dan AMCL Terhadap Degradasi Sensor dan Gangguan Lingkungan. Other thesis, Institute Teknologi Sepuluh Nopember.

Text 5022221165_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Sistem lokalisasi robot mobil non-holonomik bergantung pada akurasi dan ketahanan (robustness) estimasi pose, terutama ketika beroperasi pada lingkungan yang tidak sepenuhnya statis. Penelitian ini mengevaluasi ketahanan sistem localization mobil robot berbasis sensor fusion terhadap gangguan objek dinamis, noise sensor, dan error odometri pada lingkungan simulasi ROS2 dan Gazebo. Data wheel encoder dan IMU digabungkan menggunakan Extended Kalman Filter (EKF) untuk menghasilkan estimasi odometri, yang kemudian dikoreksi menggunakan Adaptive monte carlo Localization (AMCL) berbasis data LiDAR 2D. Sistem navigasi diimplementasikan menggunakan Navigation2 (Nav2) dengan global planner berbasis Dijkstra dan local controller berbasis DWBLocalPlanner. Hasil pengujian menunjukkan bahwa keberadaan obstacle dinamis tidak memberikan pengaruh signifikan terhadap akurasi posisi (RMSE 0,116 m, lebih rendah dari skenario obstacle statis 0,209 m dan dalam rentang baseline), namun meningkatkan deviasi orientasi hingga rata-rata 5,08°. Sistem mampu mentoleransi noise sensor LiDAR hingga σ ≈ 0,15 m dan error odometri hingga sekitar 5–8% sebelum mengalami penurunan performa signifikan, yang didefinisikan sebagai kegagalan mencapai goal atau RMSE posisi melebihi 1 m. Tingkat keberhasilan navigasi mencapai 100% pada skenario obstacle statis (S3) dan dinamis (S4), sementara skenario navigasi tanpa obstacle (S2) mencapai 80%. Hasil ini menunjukkan bahwa integrasi sensor fusion EKF dan AMCL efektif menjaga stabilitas estimasi pose pada lingkungan dinamis terbatas.
======================================================================================================================================
The performance of localization systems for nonholonomic mobile robots is closely tied to how accurately and reliably they can estimate pose, especially when the surrounding environment is not fully static. This research assesses how well a sensor fusion-based localization system for mobile robots withstands three types of disturbance moving obstacles, sensor noise, and odometry drift within a ROS2 and Gazebo simulation setting. An Extended Kalman Filter (EKF) combines wheel encoder and IMU readings to produce an odometry estimate, which is then refined through Adaptive monte carlo Localization (AMCL) using 2D LiDAR observations. Navigation is handled through Navigation2 (Nav2), employing a Dijkstra-based global planner alongside a DWB local controller. Testing revealed that dynamic obstacles had little effect on positional accuracy RMSE reached 0.116 m in the dynamic scenario (S4), actually lower than the 0.209 m recorded for the static obstacle case (S3) though orientation error rose to an average of 5.08°. The system tolerated LiDAR noise up to roughly σ ≈ 0.15 m and odometry errors around 5–8% before performance dropped markedly, with failure defined as either not reaching the goal or positional RMSE exceeding 1 m. Navigation succeeded 100% of the time in both the static (S3) and dynamic (S4) obstacle scenarios, compared to 80% in the obstacle-free case (S2). Taken together, these findings suggest that combining EKF-based sensor fusion with AMCL keeps pose estimation stable even in moderately dynamic settings

Item Type:	Thesis (Other)
Uncontrolled Keywords:	AMCL, EKF, Mobil Robot Non-Holonomik, Ketahanan, Fusi Sensor
Subjects:	T Technology > T Technology (General) > T58.8 Productivity. Efficiency T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3070 Automatic control
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Affief Adli Taqy Garninto
Date Deposited:	28 Jul 2026 00:58
Last Modified:	28 Jul 2026 00:58
URI:	http://repository.its.ac.id/id/eprint/137930

Actions (login required)

View Item