3D scene graph (3DSG) generation is a rapidly evolving field that plays a significant role in robotic autonomy. Traditionally, the focus has been on indoor environments, where robots understand and navigate spaces by abstracting objects and geometric information in a structured graph format. Expanding upon this idea, this paper introduces a 3DSG construction architecture, which enables scene-agnostic abstraction of the environment, with the goal of facilitating the adoption of 3DSG for autonomous agents in both indoor and outdoor environments. We propose a novel approach for area delimitation in 3DSGs that leverages label propagation to cluster entities (i.e. objects of interest) into areas that are both semantically and topologically distinguishable within a scene. Towards this end, we establish label propagation for 3DSGs, by formulating a dynamic set of propagation factors that accommodate to the relevance of semantic information and their natural decay through the topological structure of the 3DSG. Additionally, to achieve scene-agnostic area delimitation, we introduce a single-step optimization process for the calculation of clutter-aware propagation factors based on the approximation of an optimal set of factors that maximize inter-area eccentricity while minimizing intra-area eccentricity. Finally, the proposed framework is extensively validated through simulations and real-world deployments using a Boston Dynamics Spot legged robot and a Clearpath Husky mobile robot. The experimental results showcase the scalability of the proposed framework to indoor and outdoor environments for real-time 3DSG construction.

In recent years, autonomous robotic exploration has become an increasingly vital capability, particularly in environments where human access is limited, hazardous, or impossible. Examples include post-disaster search-and-rescue missions, subterranean cave and mine inspections, and planetary lava tube explorations. In such scenarios, autonomous systems offer not only enhanced safety but also the ability to rapidly gather critical data that would otherwise be inaccessible, improving both the efficiency and safety of field operations. As robotic systems continue to evolve, their ability to function independently in complex and GPS-denied environments relies on advances in mapping, contextual understanding of the environment and decision making. Meeting these demands requires developing frameworks that are adaptable, reliable, and capable of real-world operation beyond controlled simulation environments.

This thesis presents the development of a series of exploration and navigation frameworks designed to advance autonomous field robotics across diverse platforms, including Unmanned Aerial Vehicles (UAVs), quadrupeds, and ground-based rovers. The first contribution introduces the Rapid Exploration Framework (REF), a frontier-based system that integrates adaptive frontier switching and energy-aware decision-making to ensure safe and efficient exploration in resource-constrained subterranean settings. Building on this foundation, a rolling graph-based exploration strategy is developed, capable of incrementally constructing and updating local subgraphs in real time. This approach enables scalable and memory-efficient global navigation, maintaining continuity and computational efficiency as the environment evolves. The thesis then introduces the Semantics and Traversability-Aware Graph-based Exploration (STAGE) planner, which augments graph-based planning with real-time semantic and traversability assessments. By continuously adapting navigation graphs to reflect environmental changes, STAGE achieves reliable and collision-free motion in dynamic and uncertain terrains. The final stage of this research presents STAGE 2.0, a hierarchical and multimodal exploration system that enables cooperation between aerial and ground robots. By sharing geometric and semantic traversability information through a layered graph representation, ground robots can identify low-traversability or high-risk areas and request aerial support where necessary. This coordination allows aerial robots to extend exploration beyond physical barriers without requiring full map exchanges. The frameworks have been extensively validated through both simulation and real-world field trials in underground mines and natural cave environments, demonstrating their robustness and readiness for practical deployment. The resulting systems contribute directly to real-world applications such as autonomous mine inspection and rapid rockfall assessment, representing a significant step toward deploying intelligent, self-reliant robotic explorers capable of operating in the most demanding and unstructured environments.

In this work, we introduce SPADE, a path planning framework designed for autonomous navigation in dynamic environments using 3D scene graphs. SPADE combines hierarchical path planning with local geometric awareness to enable collision-free movement in dynamic scenes. The framework bifurcates the planning problem into two: (a) solving the sparse abstract global layer plan and (b) iterative path refinement across denser lower local layers in step with local geometric scene navigation. To ensure efficient extraction of a feasible route in a dense multi-task domain scene graphs, the framework enforces informed sampling of traversable edges prior to path-planning. This removes extraneous information not relevant to path-planning and reduces the overall planning complexity over a graph. Existing approaches address the problem of path planning over scene graphs by decoupling hierarchical and geometric path evaluation processes. Specifically, this results in an inefficient replanning over the entire scene graph when encountering path obstructions blocking the original route. In contrast, SPADE prioritizes local layer planning coupled with local geometric scene navigation, enabling navigation through dynamic scenes while maintaining efficiency in computing a traversable route. We validate SPADE through extensive simulation experiments and real-world deployment on a quadrupedal robot, demonstrating its efficacy in handling complex and dynamic scenarios. 

This work presents a fully integrated tree-based combined exploration-planning algorithm: exploration-rapidly-exploring random trees (RRT) (ERRT). The algorithm is focused on providing real-time solutions for local exploration in a fully unknown and unstructured environment while directly incorporating exploratory behavior, robot-safe path planning, and robot actuation into the central problem. ERRT provides a complete sampling and tree-based solution for evaluating “where to go next” by considering a tradeoff between maximizing information gain and minimizing the distances traveled and the robot actuation along the path. The complete scheme is evaluated in extensive simulations, comparisons, and real-world field experiments in constrained and narrow subterranean and GPS-denied environments. The framework is fully robot operating system (ROS) integrated and straightforward to use.

In this article, we propose the novel concept of Belief Scene Graphs, which are utility-driven extensions of partial 3D scene graphs, that enable efficient high-level task planning with partial information. We propose a graph-based learning methodology for the computation of belief (also referred to as expectation) on any given 3D scene graph, which is then used to strategically add new nodes (referred to as blind nodes) that are relevant to a robotic mission. We propose the method of Computation of Expectation based on Correlation Information (CECI), to reasonably approximate real Belief/Expectation, by learning histograms from available training data. A novel Graph Convolutional Neural Network (GCN) model is developed, to learn CECI from a repository of 3D scene graphs. As no database of 3D scene graphs exists for the training of the novel CECI model, we present a novel methodology for generating a 3D scene graph dataset based on semantically annotated real-life 3D spaces. The generated dataset is then utilized to train the proposed CECI model and for extensive validation of the proposed method. We establish the novel concept of \textit{Belief Scene Graphs} (BSG), as a core component to integrate expectations into abstract representations. This new concept is an evolution of the classical 3D scene graph concept and aims to enable high-level reasoning for task planning and optimization of a variety of robotics missions. The efficacy of the overall framework has been evaluated in an object search scenario, and has also been tested in a real-life experiment to emulate human common sense of unseen-objects. 

For a video of the article, showcasing the experimental demonstration, please refer to the following link: \url{https://youtu.be/hsGlSCa12iY}

As the routine operations are starting to become highly automated, it is crucial to develop autonomous solutions that are infrastructure independent. Achieving this is challenging due to the ever-changing landscape of mines, which complicates infrastructure development. In response, this paper introduces a robust framework employing drones to gather data from hard-to-access areas in mines and deliver the data back to the base station for routine monitoring purposes. These tasks include gathering data from operational vehicles (mine trucks, loaders etc.), as well as various sensors (e.g. monitoring rock bolts) and relaying the data to the mine’s base station for monitoring purposes. The proposed framework is based on autonomous navigation using a known point cloud map of the mine, proximity detection via Ultra WideBand (UWB) radios and the data transfer is accomplished through the IEEE 802.15.4 communication standard, operating in the 868 MHz ISM band, with the aim to guarantee long range operation. On the mission level, the drones act as data mules capable of autonomously extracting data from operating vehicles, storing the data onboard and eventually delivering the data to the base station, which is enabled through a Point and Click (PAC) autonomy framework based on global planning, reactive navigation, communication link and behavior management. The efficacy of this framework has been demonstrated through real-world experiments conducted at a test mine in Sweden, validating the overall architecture of the proposed solution.

This work presents EAT (Environment Agnostic Traversability for Reactive Navigation) a novel framework for traversability estimation in indoor, outdoor, subterranean (SubT) and other unstructured environments. The architecture provides updates on traversable regions online during the mission, adapts to varying environments, while being robust to noisy semantic image segmentation. The proposed framework considers terrain prioritization based on a novel decay exponential function to fuse the semantic information and geometric features extracted from RGB-D images to obtain the traversability of the scene. Moreover, EAT introduces an obstacle inflation mechanism on the traversability image, based on mean-window weighting module, allowing to adapt the proximity to untraversable regions. The overall architecture uses two LRASPP MobileNet V3 large Convolutional Neural Networks (CNN) for semantic segmentation over RGB images, where the first one classifies the terrain types and the second one classifies see-through obstacles in the scene. Additionally, the geometric features profile the underlying surface properties of the local scene, extracting normals from depth images. The proposed scheme was integrated with a control architecture in reactive navigation scenarios and was experimentally validated in indoor and outdoor environments as well as in subterranean environments with a Pioneer 3AT mobile robot.

This article studies the commonsense object affordance concept for enabling close-to-human task planning and task optimization of embodied robotic agents in urban environments. The focus of the object affordance is on reasoning how to effectively identify object’s inherent utility during the task execution, which in this work is enabled through the analysis of contextual relations of sparse information of 3D scene graphs. The proposed framework develops a Correlation Information (CECI) model to learn probability distributions using a Graph Convolutional Network, allowing to extract the commonsense affordance for individual members of a semantic class. The overall framework was experimentally validated in a real-world indoor environment, showcasing the ability of the method to level with human commonsense. For a video of the article, showcasing the experimental demonstration, please refer to the following link: https://youtu.be/BDCMVx2GiQE

In this article, we propose a novel navigation framework that leverages a two layered graph representation of the environment for efficient large-scale exploration, while it integrates a novel uncertainty awareness scheme to handle dynamic scene changes in previously explored areas. The framework is structured around a novel goal oriented graph representation, that consists of, i) the local sub-graph and ii) the global graph layer respectively. The local sub-graphs encode local volumetric gain locations as frontiers, based on the direct pointcloud visibility, allowing fast graph building and path planning. Additionally, the global graph is build in an efficient way, using node-edge information exchange only on overlapping regions of sequential sub-graphs. Different from the state-of-the-art graph based exploration methods, the proposed approach efficiently re-uses sub-graphs built in previous iterations to construct the global navigation layer. Another merit of the proposed scheme is the ability to handle scene changes (e.g. blocked pathways), adaptively updating the obstructed part of the global graph from traversable to not-traversable. This operation involved oriented sample space of a path segment in the global graph layer, while removing the respective edges from connected nodes of the global graph in cases of obstructions. As such, the exploration behavior is directing the robot to follow another route in the global re-positioning phase through path-way updates in the global graph. Finally, we showcase the performance of the method both in simulation runs as well as deployed in real-world scene involving a legged robot carrying camera and lidar sensor.