Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Zorana Milosevic has a total of 26 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Research on the use of RRTs for fast obstacle avoidance planning on unmanned wing-in-ground effect vehicles

    by Daniel Bajo Collados, Zorana Hakonen-Milošević, Sergio Domínguez Cabrerizo

    Wing-in-Ground (WIG) vehicles, also known as ekranoplans, exploit aerodynamic ground effect by flying at very low altitude over water. Operating in ground effect yields a substantially higher lift-to-drag ratio than free-flight at altitude. This translates into a vehicle capable of running faster than surface vessels and more efficiently than conventional aircraft. Their unmanned variant, Unmanned Wing-in-Ground Vehicles (UWV), is attractive for defense and security applications, where the low-altitude flight regime additionally provides a low radar signature, offering tactical advantages over conventional UAVs and USVs. This low-altitude regime also makes onboard collision avoidance non-trivial. The platform considered here is a 5.5 m fixed-wing UWV cruising at 25 m/s, with tight kinematic constraints and very limited altitude margin. With those constraints, even sparse obstacles (buoys, small vessels, debris) must be handled by fast onboard replanning. This work addresses real-time replanning for UWV under these conditions. Rapidly-exploring Random Trees (RRTs) have been successfully used for a wide range of path-planning problems, as they are probabilistically complete and require relatively low computational time. We benchmark a set of sampling-based planners derived from RRT (RRT, RRT*, RRT-Connect, LBTRRT, BgRRT and 1-0-BgRRT) on a 3-dimensional Dubins-path formulation that captures the vehicle’s turn-rate and altitude constraints. The scenario consists of a lake map with a precalculated path and an obstacle blocking it. Each planner is given a 4-second budget to avoid it, corresponding to a 100m range until reaching an unavoidable distance to the obstacle. Each configuration is repeated 100 times across multiple obstacle sizes. We report success rate, planning time and resulting path length. Results show that 1-0 BgRRT achieves 100% success, outperforming the second-best planner, BgRRT, with a 72% success rate. Other planners, such as RRT, RRT* or RRT-Connect, got success rates lower than 10%.

  • Conceptual Development of an Autonomous All-Electric Wing-in-Ground Effect Cargo Craft

    by Jussi Aaltonen, Akseli Arola, Peetu Tuominen, Kalle Hakonen-Milošević, Zorana Hakonen-Milošević, Kari Koskinen

  • Open Access

    CNN-Based Vessel Tracking in Infrared Images With an Unmanned Wing-in-Ground Vehicle

    by Miguel Domínguez, Sarthak Mishra, Ramón A. Suárez Fernández, Zorana Milošević, Claudio Rossi, Sergio Domínguez

    The urge for the development of unmanned autonomous vehicles that can satisfy the needs of the supply chains, covering distances in the shortest time possible while being environmentally sustainable, has prompted new lines of research that try to achieve all of these requirements into existing, yet not that popular vehicles, such as the wing-in-ground effect vehicle, whose proximity to the ground during its flights results in a reduction of drag and high lift, which translates into excellent performances in terms of energy consumption. Focusing on providing the vehicle with another tool towards autonomous operability, in this paper, we have developed a vessel detection and tracking system for the wing-in-ground craft using Convolutional Neural Networks trained to detect boats in infrared images. In order to test and validate the system, we have developed a dual simulator architecture using X-Plane 12 as the flight and physics simulator and Gazebo as the sensors system and the maritime environment simulator. The results of this article and the experiments presented here are a step forward toward the future operability of an unmanned wing-in-ground vehicle in defense, surveillance, and search and rescue missions performed at night or under low visibility conditions, in complex scenarios such as the maritime environment.

  • An LQG Approach to the Control of Unmanned Wing-In-Ground Vehicle

    by Sofija Ilić, Zorana Milošević, Ramon A. Suarez Fernandez, Claudio Rossi, Sergio Domínguez

    In this paper, we propose a Linear Quadratic Gaussian (LQG) approach to control an unmanned Wing-In-Ground (WIG) vehicle. We use a nonlinear and nonstationary state-space six degrees of freedom (6 DOF) model, obtained by system identification, around the trimmed operating point. Implementing the proposed control strategy includes defining the desired trajectory of motion, linearizing the model around the trim operating point, and computing adequate controller gains. The feasibility of implementing the proposed approach has been demonstrated through system behavior simulations in various conditions using Matlab and Simulink.


  • Open Access

    Topological Navigation for Autonomous Underwater Vehicles in Confined Semi-Structured Environments

    by Claudio Rossi, Adrian Caro Zapata, Zorana Milosevic, Ramon Suarez, Sergio Dominguez

    In this work, we present the design, implementation, and simulation of a topology-based navigation system for the UX-series robots, a spherical underwater vehicle designed to explore and map flooded underground mines. The objective of the robot is to navigate autonomously in the 3D network of tunnels of a semi-structured but unknown environment in order to gather geoscientific data. We start from the assumption that a topological map has been generated by a low-level perception and SLAM module in the form of a labeled graph. However, the map is subject to uncertainties and reconstruction errors that the navigation system must address. First, a distance metric is defined to compute node-matching operations. This metric is then used to enable the robot to find its position on the map and navigate it. To assess the effectiveness of the proposed approach, extensive simulations have been carried out with different randomly generated topologies and various noise rates.


  • Open Access

    Overcoming the challenges of 3D modeling in harsh, confined, underwater environments: A case study

    by Zorana Milosevic, Richard Zoltan Papp, Hilco van Moerkerk

    It is estimated that there are more than 8000 abandoned, flooded mines in Europe, many of which lack any information on their status and layout. Accurate and detailed 3D modeling plays a key role in fully understanding these complex environments and determining their remaining hidden potential. However, acquiring the needed data is a challenging task since these environments are extremely hazardous for traditional methods such as human diving. Additionally, human divers can reach only a limited depth range, much smaller than that of a standard mine. Therefore, underwater vehicles appear as a natural alternative for overcoming the disadvantages of direct human exploration. The UX1-Neo is a semi-autonomous underwater robotic system built precisely for this use. This small spherical robot with a 0.7m diameter has a 500m depth rating and various sensors for surveying the environment, such as multibeam and scanning sonars, structured light projectors, and multispectral cameras.  The unfavorable properties of the water medium, such as light scattering and attenuation, pose additional difficulties for data acquisition in these complex environments. Furthermore, mine tunnels are a GPS-denied environment, which makes the modeling system rely entirely on the robot's inertial navigation system, which is prone to error due to the dead-reckoning drift. Conventional methods for correcting this drift, such as SLAM, face additional challenges in these repetitive environments (shafts and tunnels) due to their highly symmetric structures and lack of distinctive features. Additionally, during the exploration of a salt mine, Solotvyno (Ukraine), we faced a new challenge, a refraction of the sonar data due to the salty water, which required further processing in order to create an accurate 3D map of the mine.  Rapid developments in the field of underwater photogrammetry are producing impressive results; however, they still have difficulties with the environments with low light, which causes blurring of details, low image contrast, and in general, lack of features needed for image matching. Also, underwater images are prone to contain an excessive amount of blue light, making the features even less visible. Moreover, photogrammetry technology struggles with repetitive environments due to the same reasons as SLAM. In this work, we demonstrate the challenges faced during our exploration of the Solotvyno salt mine with the UX1-Neo robot and how we overcame them in order to produce a detailed 3D model. In particular, we illustrate that sensor and data redundancy is vital during operations and post-processing. Each UX1-Neo sensor contributes to a complete, coherent picture of the environment. However, using many sensors produces an enormous amount of data that require further filtering: hundreds of millions of points are reduced to a few million using both automated and manual methods. Images also require processing due to the aforementioned reasons: using CLAHE contrast enhancement together with white balancing algorithms, we produce suitable images for photogrammetry. Additionally, data gathered from multiple missions need to be combined for a complete model: we show the importance of robot orientation initialization and external surveying of the robot's launch location to correctly align scans of different missions.

  • Open Access

    Software-in-the-loop and hardware-in-the-loop paradigms and their use for research and development of autonomous underwater vehicles

    by Zorana Milosevic, Ramon A. Suarez Fernandez, Sergio Dominguez, Claudio Rossi, Richard Zoltan Papp, Hilco van Moerkerk

    The development and field testing of autonomous robots are complex tasks for a number of reasons. The involved logistics are often quite complicated, and the risk of damaging the equipment under evaluation is very high; this could, on occasion, represent a considerable time and monetary overload. Such difficulties are greatly magnified in the development and testing of the underwater platforms designed to operate in hazardous environments, such as underground flooded mines, the main focus of the UNEXUP project, funded under EIT Raw Materials, and its predecessor, the horizon 2020 UNEXMIN project. These unique field working conditions involve a high risk of permanent loss of the platform in case of failure or error and a high risk of casualties in rescue attempts entailing direct human actions. In contrast, software-in-the-loop (SIL) and hardware-in-the-loop (HIL) testing paradigms are powerful tools for preliminary system validation and algorithm benchmarking where troubleshooting is simplified through the use of a controlled environment. They provide a time- and cost-effective solution for testing, thus having a crucial role in the development of advanced autonomous robotic platforms. These two paradigms are possible thanks to the evolution of simulation models, which have achieved astonishing completeness and can realistically simulate not only system dynamics but also all the operational components of a robotic platform, such as their sensor readings, even corrupted with realistic noise. SIL experiments involve software uniquely, and are thus performed without the real robot platform but its simulation model, making them an ideal tool for testing specific algorithms or software modules. On the other hand, HIL experiments involve the real robot’s hardware, either the complete robotic platform or only parts of it, thus providing a more realistic testing environment. In this work, we illustrate the vast range of aspects during the development of a robotic platform that can dramatically benefit from the use of the combination of SIL and HIL testing, especially in those technical applications where field trials show severe operational difficulties. We show how these testing paradigms provide a solid basis for evaluating parts or modules of a system, which, besides being convenient for the development of advanced robotic platforms by multiple teams, also bridges the gap between algorithm design and the testing of a complete platform. Then, we show how SIL and HIL can substitute parts of real environments, and enrich aspects of real data to focus on specific testing situations not easily controllable, or even dangerous, in field tests. We demonstrate how we can create augmented environments by introducing virtual obstacles and even complete virtual maps into the available experimental setup, such as a real submersible inside a water tank, thus testing complex maneuvers and simulating possible real scenarios with minimum risk of compromising the equipment. We show how these environments are beneficial not only when developing autonomous submersibles but also when training human operators of non-autonomous ones in a so-called human-in-the-loop configuration. 

  • Open Access

    Functional Self-Awareness and Metacontrol for Underwater Robot Autonomy

    by Esther Aguado, Zorana Milosevic, Carlos Hernández, Ricardo Sanz, Mario Garzon, Darko Bozhinoski, Claudio Rossi

    Autonomous systems are expected to maintain a dependable operation without human intervention. They are intended to fulfill the mission for which they were deployed, properly handling the disturbances that may affect them. Underwater robots, such as the UX-1 mine explorer developed in the UNEXMIN project, are paradigmatic examples of this need. Underwater robots are affected by both external and internal disturbances that hamper their capability for autonomous operation. Long-term autonomy requires not only the capability of perceiving and properly acting in open environments but also a sufficient degree of robustness and resilience so as to maintain and recover the operational functionality of the system when disturbed by unexpected events. In this article, we analyze the operational conditions for autonomous underwater robots with a special emphasis on the UX-1 miner explorer. We then describe a knowledge-based self-awareness and metacontrol subsystem that enables the autonomous reconfiguration of the robot subsystems to keep mission-oriented capability. This resilience augmenting solution is based on the deep modeling of the functional architecture of the autonomous robot in combination with ontological reasoning to allow self-diagnosis and reconfiguration during operation. This mechanism can transparently use robot functional redundancy to ensure mission satisfaction, even in the presence of faults.


  • Open Access

    Guidance for Autonomous Underwater Vehicles in Confined Semistructured Environments

    by Zorana Milosevic, Ramon A. Suarez Fernandez, Sergio Dominguez, Claudio Rossi

    In this work, we present the design, implementation, and testing of a guidance system for the UX-1 robot, a novel spherical underwater vehicle designed to explore and map flooded underground mines. For this purpose, it needs to navigate completely autonomously, as no communications are possible, in the 3D networks of tunnels of semistructured but unknown environments and gather various geoscientific data. First, the overall design concepts of the robot are presented. Then, the guidance system and its subsystems are explained. Finally, the system’s validation and integration with the rest of the UX-1 robot systems are presented. A series of experimental tests following the software-in-the-loop and the hardware-in-the-loop paradigms have been carried out, designed to simulate as closely as possible navigation in mine tunnel environments. The results obtained in these tests demonstrate the effectiveness of the guidance system and its proper integration with the rest of the systems of the robot, and validate the abilities of the UX-1 platform to perform complex missions in flooded mine environments.


  • Open Access

    Motion Control of Underwater Mine Explorer Robot UX-1: Field Trials

    by Ramon A. Suarez Fernandez, Zorana Milosevic, Sergio Dominguez, Claudio Rossi

    In this work, we present the results of a series of field trials conducted with an underwater vehicle in the increasingly complex underwater environments of the Kaatiala mine in Finland, the Idrija mine in Slovenia, and the Urgeiriça mine in Portugal. These field experiments have been performed to test and validate the motion control systems of the underwater explorer robot (UX-1), which are: a variable ballast system for buoyancy control, a variable pitch system for pitch control, and a propulsion system for depth and heading stabilization. The control method implemented is based on a state Feedback Linearization algorithm, previously tested and validated in a controlled water tank environment. Several experiments are shown which demonstrate the successful operation of the motion control systems in individual performance tests. Afterwards, a full exploration and geo-scientific data collection mission is presented, where all of the motion control systems are active simultaneously during a deep dive to more than 100 m. The results obtained in the field tests demonstrate the effectiveness of the motion control systems and validate the UX-1 Robot platform as capable of performing the complex task of navigation and control in flooded mine environments.