Collaborative multi-agent exploration of unknown environments is crucial for search and rescue operations. Effective real-world deployment must address challenges such as limited inter-agent communication and static and dynamic obstacles. This paper introduces a novel decentralized collaborative framework based on Reinforcement Learning to enhance multi-agent exploration in unknown environments. Our approach enables agents to decide their next action using an agent-centered field-of-view occupancy grid, and features extracted from A* algorithm-based trajectories to frontiers in the reconstructed global map. Furthermore, we propose a constrained communication scheme that enables agents to share their environmental knowledge efficiently, minimizing exploration redundancy. The decentralized nature of our framework ensures that each agent operates autonomously, while contributing to a collective exploration mission. Extensive simulations in Gymnasium and real-world experiments demonstrate the robustness and effectiveness of our system, while all the results highlight the benefits of combining autonomous exploration with inter-agent map sharing, advancing the development of scalable and resilient robotic exploration systems.

Autonomous navigation in unknown environments is a fundamental challenge in robotics, particularly in coordinating ground and aerial robots to maximize exploration efficiency. This paper presents a novel approach that utilizes a hierarchical graph to represent the environment, encoding both geometric and semantic traversability. The framework enables the robots to compute a shared confidence metric, which helps the ground robot assess terrain and determine when deploying the aerial robot will extend exploration. The robot's confidence in traversing a path is based on factors such as predicted volumetric gain, path traversability, and collision risk. A hierarchy of graphs is used to maintain an efficient representation of traversability and frontier information through multi-resolution maps. Evaluated in a real subterranean exploration scenario, the approach allows the ground robot to autonomously identify zones that are no longer traversable but suitable for aerial deployment. By leveraging this hierarchical structure, the ground robot can selectively share graph information on confidence-assessed frontier targets from parts of the scene, enabling the aerial robot to navigate beyond obstacles and continue exploration.

This work establishes the concept of commonsense scene composition, with a focus on extending Belief Scene Graphs by estimating the spatial distribution of unseen objects. Specifically, the commonsense scene composition capability refers to the understanding of the spatial relationships among related objects in the scene, which in this article is modeled as a joint probability distribution for all possible locations of the semantic object class. The proposed framework includes two variants of a Correlation Information (CECI) model for learning probability distributions: (i) a baseline approach based on a Graph Convolutional Network, and (ii) a neuro-symbolic extension that integrates a spatial ontology based on Large Language Models (LLMs). Furthermore, this article provides a detailed description of the dataset generation process for such tasks. Finally, the framework has been validated through multiple runs on simulated data, as well as in a real-world indoor environment, demonstrating its ability to spatially interpret scenes across different room types. For a video of the article, showcasing the experimental demonstration, please refer to the following link: https://youtu.be/f0tqtPVFZ2A

In this article, we address the problem of collaborative task assignment, sequencing, and multi-agent pathfinding (TSPF), where a team of agents must visit a set of task locations without collisions while minimizing flowtime. TSPF incorporates agent-task compatibility constraints and ensures that all tasks are completed. We propose a Conflict-Based Search with Task Sequencing (CBS-TS), an optimal and complete algorithm that alternates between finding new task sequences and resolving conflicts in the paths of current sequences. CBS-TS uses a mixed-integer linear program (MILP) to optimize task sequencing and employs Conflict-Based Search (CBS) with Multi-Label A* (MLA*) for collision-free path planning within a search forest. By invoking MILP for the next-best sequence only when needed, CBS-TS efficiently limits the search space, enhancing computational efficiency while maintaining optimality.We compare the performance of our CBS-TS against Conflict-based Steiner Search (CBSS), a baseline method that, with minor modifications, can address the TSPF problem. Experimental results demonstrate that CBS-TS outperforms CBSS in most testing scenarios, achieving higher success rates and consistently optimal solutions, whereas CBSS achieves near-optimal solutions in some cases. The supplementary video is available at https://youtu.be/QT8BYgvefmU.

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.

In this article, we present the Layered Semantic Graphs (LSG), a novel actionable hierarchical scene graph, fully integrated with a multi-modal mission planner, the FLIE: A First-Look based Inspection and Exploration planner [1]. The novelty of this work stems from aiming to address the task of maintaining an intuitive and multi-resolution scene representation, while simultaneously offering a tractable foundation for planning and scene understanding during an ongoing inspection mission of apriori unknown targets-of-interest in an unknown environment. The proposed LSG scheme is composed of locally nested hierarchical graphs, at multiple layers of abstraction, with the abstract concepts grounded on the functionality of the integrated FLIE planner. Furthermore, LSG encapsulates real-time semantic segmentation models that offer extraction and localization of desired semantic elements within the hierarchical representation. This extends the capability of the inspection planner, which can then leverage LSG to make an informed decision to inspect a particular semantic of interest. We also emphasize the hierarchical and semantic path-planning capabilities of LSG, which could extend inspection missions by improving situational awareness for human operators in an unknown environment. The validity of the proposed scheme is proven through extensive evaluations of the proposed architecture in simulations, as well as experimental field deployments on a Boston Dynamics Spot quadruped robot in urban outdoor environment settings.

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.

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 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: https://youtu.be/hsGlSCa12iY

In this article, we propose a novel communication-based action space enhancement for the D-MARL exploration algorithm to improve the efficiency of mapping an unknown environment, represented by an occupancy grid map. In general, communication between autonomous systems is crucial when exploring large and unstructured environments. In such real-world scenarios, data transmission is limited and relies heavily on inter-agent proximity and the attributes of the autonomous platforms. In the proposed approach, each agent’s policy is optimized by utilizing the heterogeneous-agent proximal policy optimization algorithm to autonomously choose whether to communicate or explore the environment. To accomplish this, multiple novel reward functions are formulated by integrating inter-agent communication and exploration. The investigated approach aims to increase efficiency and robustness in the mapping process, minimize exploration overlap, and prevent agent collisions. The D-MARL policies trained on different reward functions have been compared to understand the effect of different reward terms on the collaborative attitude of the homogeneous agents. Finally, multiple simulation results are provided to prove the efficacy of the proposed scheme.

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

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 establishes the novel D+∗, a risk-aware and platform-agnostic heterogeneous global path planner for robotic navigation in complex environments. The proposed planner addresses a fundamental bottleneck of occupancy-based path planners related to their dependency on accurate and dense maps. More specifically, their performance is highly affected by poorly reconstructed or sparse areas (e.g. holes in the walls or ceilings) leading to faulty generated paths outside the physical boundaries of the 3-dimensional space. As it will be presented, D+∗ addresses this challenge with three novel contributions, integrated into one solution, namely: (a) the proximity risk, (b) the modeling of the unknown space, and (c) the map updates. By adding a risk layer to spaces that are closer to the occupied ones, some holes are filled, and thus the problematic short-cutting through them to the final goal is prevented. The novel established D+∗ also provides safety marginals to the walls and other obstacles, a property that results in paths that do not cut the corners that could potentially disrupt the platform operation. D+∗ has also the capability to model the unknown space as risk-free areas that should keep the paths inside, e.g in a tunnel environment, and thus heavily reducing the risk of larger shortcuts through openings in the walls. D+∗ is also introducing a dynamic map handling capability that continuously updates with the latest information acquired during the map building process, allowing the planner to use constant map growth and resolve cases of planning over outdated sparser map reconstructions. The proposed path planner is also capable to plan 2D and 3D paths by only changing the input map to a 2D or 3D map and it is independent of the dynamics of the robotic platform. The efficiency of the proposed scheme is experimentally evaluated in multiple real-life experiments where D+∗ is producing successfully proper planned paths, either in 2D in the use case of the Boston dynamics Spot robot or 3D paths in the case of an unmanned areal vehicle in varying and challenging scenarios.

Robust localization is a fundamental capability to increase the autonomy levels of robotic platforms. A core processing step in vision based odometry methods is the extraction and tracking of distinctive features in the image frame. Nevertheless, when deploying robots in challenging environments like underground tunnels, the sensor measurements are noisy with lack of information due to low light conditions, introducing a bottleneck for feature detection methods. This paper proposes a deep classifier Convolutional Neural Network (CNN) architecture to retain detailed and noise tolerant feature maps from RBG images, establishing a novel feature tracking scheme in the context of localization. The proposed method is feeding the RGB image into the AlexNet or VGG-16 network and extracts a feature map at a specific layer. This feature map consists of feature points which are then paired between frames resulting in a discrete vector field of feature change. Finally, the proposed method is evaluated with RGB camera footage of the Micro Aerial Vehicle (MAV) flights in dark underground mines and the performance is compared with existing feature extraction methods, while the noise is added to the images.

Degraded Subterranean environments are an attractive case for miniature aerial vehicles, since there is a constant need to increase the safety operations in underground mines. The starting point for integrating aerial vehicles in the mining process is the capability to reliably navigate along tunnels. Inspired by recent advancements, this paper presents a collection of different, experimentally verified, methods tackling the problem of MAVs heading regulation while navigating in dark and textureless tunnel areas. More specifically, four different methods are presented in this work with the common goal to identify open space in the tunnel and align the MAV heading using either visual sensor in methods a) single image depth estimation, b) darkness contour detection, c) Convolutional Neural Network (CNN) regression and 2D Lidar sensor in method d) range geometry. For the works a)‐c) the dark scene in the middle of the tunnel is considered as open space and is processed and converted to yaw rate command, while d) examines the geometry of the range measurements to calculate the yaw rate command. Experimental results from real underground tunnel demonstrate the performance of the methods in the field, while setting the ground for further developments in the aerial robotics community.

This article proposes a novel visual framework for detecting tunnel crossings/junctions in underground mine areas towards the autonomous navigation of Micro Aerial Vehicles (MAVs). Usually mine environments have complex geometries, including multiple crossings with different tunnels that challenge the autonomous planning of aerial robots. Towards the envisioned scenario of autonomous or semi-autonomous deployment of MAVs with limited Line-of-Sight in subterranean environments, the proposed module acknowledges the existence of junctions by providing crucial information to the autonomy and planning layers of the aerial vehicle. The capability for a junction detection is necessary in the majority of mission scenarios, including unknown area exploration, known area inspection and robot homing missions. The proposed novel method has the ability to feed the image stream from the vehicles on-board forward facing camera in a Convolutional Neural Network (CNN) classification architecture, expressed in four categories: 1) left junction, 2) right junction, 3) left & right junction, and 4) no junction in the local vicinity of the vehicle. The core contribution stems for the incorporation of AlexNet in a transfer learning scheme for detecting multiple branches in a subterranean environment. The validity of the proposed method has been validated through multiple data-sets collected from real underground environments, demonstrating the performance and merits of the proposed module.

This article presents a Convolutional Neural Network (CNN) method to enable autonomous navigation of low-cost Micro Aerial Vehicle (MAV) platforms along dark underground mine environments. The proposed CNN component provides online heading rate commands for the MAV by utilising the image stream from the on-board camera, thus allowing the platform to follow a collision-free path along the tunnel axis. A novel part of the developed method consists of the generation of the data-set used for training the CNN. More specifically, inspired from single image haze removal algorithms, various image data-sets collected from real tunnel environments have been processed offline to provide an estimation of the depth information of the scene, where ground truth is not available. The calculated depth map is used to extract the open space in the tunnel, expressed through the area centroid and is finally provided in the training of the CNN. The method considers the MAV as a floating object, thus accurate pose estimation is not required. Finally, the capability of the proposed method has been successfully experimentally evaluated in field trials in an underground mine in Sweden.

This work presents an aerial tool towards the autonomous cooperative coverage and inspection of a large scale 3D infrastructure using multiple Unmanned Aerial Vehicles (UAVs). In the presented approach the UAVs are relying only on their onboard computer and sensory system, deployed for inspection of the 3D structure. In this application each agent covers a different part of the scene autonomously, while avoiding collisions. The autonomous navigation of each platform on the designed path is enabled by the localization system that fuses Ultra Wideband with inertial measurements through an Error- State Kalman Filter. The visual information collected from the aerial team is collaboratively processed to create the 3D model. The performance of the overall setup has been experimentally evaluated in realistic wind turbine inspection experiments, providing dense 3D reconstruction of the inspected structures.

This article presents an integrated vision-based guiding system for aerial manipulation. More specifically, a 4 DoF planar dexterous manipulator, with a stereo camera attached on the end-effector, is endowed to a multirotor aerial platform enabling active manipulation capabilities. The proposed novel approach combines a visual processing scheme for object detection and tracking, as well as a manipulator positioning for allowing the aerial platform to approach the surface of interaction efficiently. In the developed scheme, the object detection is based on correlation filters to track the target robustly, while the depth information, from the stereo camera on board the manipulator, is used to extract the centroid of the manipulated object, compute its relative configuration with respect to the UAV and align the end-effector properly with the grasping point. The effectiveness of the proposed scheme is demonstrated in multiple experimental trials and simulations, highlighting it's applicability towards autonomous aerial manipulation.

In this article, we propose a planning algorithm for coverage of complex structures with a network of robotic sensing agents, with multi-robot surveillance missions as our main motivating application. The sensors are deployed to monitor the external surface of a 3D structure. The algorithm controls the motion of each sensor so that a measure of the collective coverage attained by the network is nondecreasing, while the sensors converge to an equilibrium configuration. A modified version of the algorithm is also provided to introduce collision avoidance properties. The effectiveness of the algorithm is demonstrated in a simulation and validated experimentally by executing the planned paths on an aerial robot.