Autonomous Navigation of Service Robots in Complex Industrial Environments with SAR-Based Vision and Advanced Detection for Industry 4.0

Abstract

In the era of Industry 4.0 and Smart Industry, the integration of autonomous systems into industrial operations offers promising advancements in safety, efficiency, and productivity. This paper presents a novel methodology for navigating service robots, specifically unmanned ground vehicles (UGVs), within challenging industrial environments where visibility is often limited, and obstacles are abundant. Our approach employs a combination of synthetic aperture radar (SAR) simulation and dark channel processing to enhance low-light images, followed by edge extraction to accurately distinguish ground and horizon levels. The methodology further integrates the YOLO11 algorithm to identify and remove obstacles, enabling the UGV to focus on navigable paths. Finally, we apply a random point assignment and linear regression approach to generate a potential route, guiding the service robot through unstructured and dynamic industrial landscapes. The proposed method was tested on a dataset of 591 images captured in simulated industrial settings, revealing high accuracy in route generation with minimal deviation from ground-truth paths. The YOLO11 algorithm achieved a precision of 90.3% in identifying and excluding obstacles, ensuring safe and efficient path planning. The findings indicate that this methodology effectively addresses the navigational challenges faced by service robots in complex environments, aligning with the goals of Industry 4.0 by advancing the autonomy and operational resilience of industrial robotic systems.