Tunnel Exploration

Important

A Master Thesis project on the investigation and usage of tunnel exploration strategies for mobile robots.

Tunnel inspection represents a significant sector necessitating thorough surveys to validate the structural integrity and assess the infrastructure contained within, such as gas pipes, sewage systems, and electrical cables. The inherent challenges in tunnel inspection stem from their extensive and intricate networks, often obscured in darkness, compounded by the absence of network connectivity and potential hazards arising from inadequate ventilation. Real-world inspections can be time-consuming, requiring several hours to traverse just a few kilometers of tunnel. In response to these challenges, autonomous exploration using mobile robots has emerged as a promising solution, focusing on efficient data collection and navigation strategies.

One such strategy, termed Priority-Based Depth-First Search (PDFS), has been developed to autonomously explore, map, and identify key features within tunnels. Leveraging various information sources including Lidar data and odometry, PDFS addresses the complexities of subterranean environments, facilitating environment segmentation and junction identification. Comparative evaluations against established approaches like Rapidly Exploring Random Trees (RRT) conducted in both simulated (Clearpath Jackal & Turtlebot3) and real-world (Turtlebot3) scenarios demonstrate PDFS’s efficacy, particularly in supervised and data-collection-based exploration contexts. However, RRT exhibits superior speed and efficiency in exploration, highlighting the nuanced trade-offs inherent in autonomous tunnel exploration strategies. This research holds significance as it lays the groundwork for advancements in techniques related to junction detection, key location identification, and supervised exploration methodologies, offering valuable insights for further development in subterranean exploration endeavors.