Szczegóły publikacji
Opis bibliograficzny
Improving LiDAR-based SLAM accuracy using IMU and odometry sensor fusion / Raoof A., GIERGIEL M. // W: "Problemi ta ìnnovacìï u rozvitku ìnženerìï, tehnologìj ta transportu" [Dokument elektroniczny] : III mižnarodna naukova konferencìâ studentìv ì molodih včenih : April 23–25, 2026, Khmelnytskyi, [Ukraïna] = "Problems and innovations in the development of engineering, technologies and transport" : III international scientific conference of students and young scientists. — Wersja do Windows. — Dane tekstowe. — Khmelnytskyi : Khmelnytskyi National University KhNU, 2026. — e-ISBN: 966-1502-35-5. — S. 506–511. — Wymagania systemowe: Adobe Reader. — Tryb dostępu: https://s.agh.edu.pl/6atPH [2026-07-13]. — Bibliogr. s. 511, Abstr.
Autorzy (2)
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 168997 |
|---|---|
| Data dodania do BaDAP | 2026-09-01 |
| Rok publikacji | 2026 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
Abstract
This Article explores the enhancement of Simultaneous Localization and Mapping (SLAM) accuracy by integrating a 6-axis Inertial Measurement Unit (IMU) with traditional motor odometry via an Extended Kalman Filter (EKF). While LiDAR-based SLAM systems on resource-constrained platforms often suffer from cumulative drift and structural artifacts like "double-walling" due to wheel slip, this research demonstrates that probabilistic sensor fusion can effectively mitigate these issues. The methodology utilizes a Raspberry Pi 5 as the primary compute unit and an ESP32-based controller for real-time sensor data acquisition. By employing a highperformance 3D kinematic model within the ROS 2 ecosystem, the system "pins" the robot's heading to the gyroscopic data, thereby neutralizing the impact of mechanical traction loss. Experimental results, gathered through standardized "Figure-Eight" trajectories in a controlled laboratory environment, indicate that the fused EKF state estimate significantly improves geometric sharpness and topological consistency compared to encoderonly localization. This study validates that mathematical sensor fusion can bridge the performance gap between low-cost hardware and industrial-grade mapping requirements without exceeding the computational limits of embedded edge devices.
