Sensor Fusion Techniques for Autonomous Robotic Navigation in Electrical Plants

Abstract

In this paper, the researcher examine how autonomous robotic navigation in electrical plants can be made possible by sensor fusion techniques; GPS-denied conditions, electromagnetic interference, complex layout make this a difficult task. The main objective is to gauge the benefits of incorporating the various sensing modalities LiDAR, vision systems, inertial measurement units (IMUs), and wheel odometry to enhance the localization precision, obstacle detection and reliable functioning. The methodology of the research is that of secondary data, using peer-reviewed articles, IEEE conference proceeding papers, and industrial case reports in order to examine established models and proven metrics. Current results show that a multi-sensor fusion can enable centimeter-level localization accuracy, 92% obstacle recognition rates and greater than 95% successful task completion in a complex environment. Both probabilistic algorithms and deep learning fusion models are effective in terms of robustness but combining the two approaches produces the most adaptive solutions. All in all, the field study has proved the effectiveness and validity of sensor fusion in ensuring safe, accurate, and reliable in autonomous robotic activities in the electrical plants setting.