A Study on the Path Estimation of the Autonomous Mobile Robot in Extreme Environment Using Image Processing and Kalman Filter

It has been widely reported that underground power transmission tunnels are recognized for their construction significance in mitigating blackout risks caused by natural disasters and addressing concerns in narrow and densely populated urban areas. Consequently, the development of technologies, such as automatic inspection robots, plays a vital role in ensuring the safe and efficient management of these underground facilities [1]–[5]. However, various challenges arise, including network disconnections, disabled GPS functionality, magnetometer errors induced by high-voltage environments, and the presence of unpredictable variables such as brightness, moisture, and aging effects. These factors collectively impede the accurate recognition of guidance lines where docking stations installed within specific underground power transmission tunnels. This paper presents a method of path estimation utilizing the Kalman filter on guidance lines extracted through image processing. The primary objective of this approach is to supplement the autonomous navigation capabilities of mobile robots operating in areas where conventional LiDAR or camera-based autonomous patrol methods are rendered impractical due to extreme environmental conditions.