Reduksi Data Point cloud dan Navigasi Obstacle Avoidance Menggunakan Voxel grid filter dan Artificial Potential Field

Nurhadi, Aqil Rabbani (2024) Reduksi Data Point cloud dan Navigasi Obstacle Avoidance Menggunakan Voxel grid filter dan Artificial Potential Field. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022222026_Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Studi ini memperkenalkan pendekatan inovatif pada bidang lokalisasi dan pemetaan dengan mengintegrasikan sensor-sensor ke dalam robot bergerak (mobile robot) diferensial non-holonomic berkenaan dengan permasalahan dari pemetaan ialah ter-representasinya peta secara akurat dan beban komputasi yang berat baik itu ketika memperbarui pose, peta ataupun pemrosesan sensor-sensor. Pendekatan yang disarankan secara harmonis menggabungkan Simultaneous Localization and Mapping (SLAM) secara real-time dengan Voxel Grid Filter untuk mengurangi sampel data point cloud, sehingga mencapai keseimbangan ideal antara menghasilkan peta yang akurat secara dimensi dan mematuhi batasan komputasi sistem. Alih-alih memproses setiap titik data individual, sistem menghitung satu titik perwakilan untuk setiap voxel dengan menggabungkan titik-titik yang berdekatan menjadi unit voxel yang lebih besar (piksel volumetrik) menggunakan pusat massa atau posisi rata-rata semua titik dalam voxel tersebut. Sistem ini secara efisien meminimalkan tuntutan komputasi sambil mempertahankan model sekitar yang representatif melalui titik berbasis voxel. Hasil simulasi menunjukkan keberhasilan pendekatan ini dalam mengatasi tantangan terkait SLAM, khususnya dalam hal akurasi peta dan efisiensi komputasi. Memanfaatkan filter voxel grid 0.01 secara signifikan meningkatkan efisiensi iterasi, mengurangi waktu komputasi rata-rata per eksekusi algoritma menjadi 0.160 detik. Sebaliknya, pemrosesan data LiDAR mentah memerlukan 0.247 detik per iterasi, sehingga menghasilkan pengurangan waktu komputasi sebesar 35%. Peningkatan efisiensi ini dicapai tanpa mengorbankan akurasi dari peta yang dihasilkan atau ketepatan lokalisasi robot, dan dengan navigasi Artificial Potential Field, mobile robot dapat menghindari rintangan baik statis ataupun dinamis.
==============================================================================================
This study introduces an innovative approach to localization and mapping by integrating sensors into a non-holonomic differential mobile robot due to the mapping challenges of accurate map representation and heavy computational burdens in both pose, map and sensor processing updates. The proposed approach harmoniously combines real-time Simultaneous Localization and Mapping (SLAM) with Voxel Grid Filter to downsample point cloud data, achieving an ideal balance between generating dimensionally accurate maps and complying with system computational constraints. Instead of processing each individual data point, the system computes a representative point for each voxel by merging adjacent points into a larger voxel unit (volumetric pixel) using the centroid or average position of all points within that voxel. The system efficiently minimizes the computational demands while maintaining a representative model of the surroundings through voxel-based points. Simulation results demonstrate the efficacy of this approach in overcoming the challenges associated with SLAM, particularly in terms of map accuracy and computational efficiency. Utilizing a 0.01 voxel grid filter significantly improves iteration efficiency, reducing the average computation time per algorithm execution to 0.160 seconds. In contrast, processing the raw LiDAR data requires 0.247 seconds per iteration, resulting in a 35% reduction in computation time. This efficiency gain is achieved without sacrificing the accuracy of the resulting map or the robot’s localization accuracy, and with Artificial Potential Field navigation, the mobile robot can avoid both static and dynamic obstacles.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Artificial Potential Field, LiDAR 3D, Mobile robot, SLAM 3D, Voxel Grid Filter, Artificial Potential Field, LiDAR 3D, Mobile robot, SLAM 3D, Voxel Grid Filter .
Subjects:	T Technology > T Technology (General) > T385 Visualization--Technique T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T57.62 Simulation T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3070 Automatic control T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK6592.A9 Automatic tracking.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Aqil Rabbani Nurhadi
Date Deposited:	07 Aug 2024 07:41
Last Modified:	07 Aug 2024 07:41
URI:	http://repository.its.ac.id/id/eprint/111866

Actions (login required)

View Item