Current global re-localization algorithms are built on top of localization and mapping methods andheavily rely on scan matching and direct point cloud feature extraction and therefore are vulnerable infeatureless demanding environments like caves and tunnels. In this article, we propose a novel globalre-localization framework that: a) does not require an initial guess, like most methods do, while b)it has the capability to offer the top-kcandidates to choose from and last but not least provides anevent-based re-localization trigger module for enabling, and c) supporting completely autonomousrobotic missions. With the focus on subterranean environments with low features, we opt to usedescriptors based on range images from 3D LiDAR scans in order to maintain the depth informationof the environment. In our novel approach, we make use of a state-of-the-art data-driven descriptorextraction framework for place recognition and orientation regression and enhance it with the additionof a junction detection module that also utilizes the descriptors for classification purposes.