1,721,337 research outputs found
Low-thrust under-actuated satellite formation guidance and control strategies
Full text linkpeer reviewedThis study presents autonomous guidance and control strategies for reconfiguring close-range multi-satellite formations. The formation consists of N under-actuated deputy satellites and an uncontrolled virtual or physical chief spacecraft. Each deputy is equipped with a single throttleable but ungimbaled low-thrust nozzle, requiring a combination of thrust and coast arcs, during the latter attitude adjustments redirect the nozzle to the desired thrust direction. The guidance problem is formulated as a trajectory optimization task incorporating dynamical and physical constraints, along with a minimum acceleration threshold imposed by typical electric thrusters. Two frameworks are considered: centralized and distributed. The centralized approach ensures fuel-optimal solutions but is feasible only for small formations, with all calculations performed on a physical chief satellite. The distributed approach, while sub-optimal, scales better by treating the chief as a virtual point mass and allowing each deputy to handle its own computations. This study focuses on spaceborne closed-loop control implementation, ensuring reliability and automation in solving the optimal control problem. To mitigate infeasibility risks, constraints that pose potential threats are identified and softened. Two Model Predictive Control architectures, shrinking-horizon and fixed-horizon, are implemented and compared in terms of fuel consumption and control accuracy. Their performance is analyzed for typical close-range reconfigurations required in Earth observation missions and benchmarked against existing approaches in the literature
Delta-V-optimal centralized guidance strategy for under-actuated N-satellite formations
Full text linkpeer reviewedThis paper addresses the computation of Delta-V-optimal, safe, relative orbit reconfigurations for satellite formations in a centralized fashion. The formations under consideration comprise an uncontrolled chief spacecraft flying with an arbitrary number, N, of deputy satellites, where each deputy is equipped with a single electric thruster. Indeed, this represents a technological solution that is becoming widely employed by the producers of small-satellite platforms. While adopting a single electric thruster does reduce the required power, weight, and size of the orbit control system, it comes at the cost of rendering the satellite under-actuated. In this setting, the satellite can provide a desired thrust vector only after an attitude maneuver is carried out to redirect the thruster nozzle opposite to the desired thrust direction. In order to further extend the applicability range of such under-actuated platforms, guidance strategies are developed to support different reconfiguration scenarios for N-satellite formations. This paper starts from a classical non-convex quadratically constrained trajectory optimization formulation, which passes through multiple simplifications and approximations to arrive to two novel convex formulations, namely a second-order cone programming formulation, and a linear programming one. Out of five guidance formulations proposed in this article, the most promising three were compared through an extensive benchmark analysis that is applied to fifteen of the most widely-used solvers. This benchmark experiment provides information about the key distinctions between the different problem formulations, and under which conditions each one of them can be recommended.Development Tool For Autonomous Constellation And Formation Control Of Microsatellites - AuFoSa
Fuel-Optimal Formation Reconfiguration by Means of Unidirectional Low-Thrust Propulsion System
Full text linkpeer reviewedR-AGR-3763 - BRIDGES/19/MS/14302465/AuFoSat/LuxSpace - VOOS Holge
Robust Real-time Sense-and-Avoid Solutions for Remotely Piloted Quadrotor UAVs in Complex Environments
Full text linkUAV teleoperation is a demanding task: to successfully accomplish the mission without collision requires skills and experience. In real-life environments, current commercial UAVs are to a large extent remotely piloted by amateur human pilots. Due to lack of teleoperation experience or skills, they often drive UAVs into collision. Therefore, in order to ensure safety of the UAV as well as its surroundings, it is necessary for the UAV to boast the capability of detecting emergency situation and acting on its own when facing imminent threat. However, the majority of UAVs currently available in the market are not equipped with such capability. To fill in the gap, in this work we present 2D LIDAR based Sense-and-Avoid solutions which are able to actively assist unskilled human operator in obstacle avoidance, so that the operator can focus on high-level decisions and global objectives in UAV applications such as search and rescue, farming etc. Specifically, with our novel 2D LIDAR based obstacle detection and tracking algorithm, perception-assistive flight control design, progressive emergency evaluation policies and optimization based and adaptive virtual cushion force field (AVCFF) based avoidance strategies, our proposed UAV teleoperation assistance systems are capable of obstacle detection and tracking, as well as automatic obstacle avoidance in complex environment where both static and dynamic objects are present. Additionally, while the optimization based solution is validated in Matlab, the AVCFF based avoidance system has been fully integrated with sensing system, perception-assistive flight controller on the basis of the Hector Quadrotor open source framework, and the effectiveness of the complete Sense-and-Avoid solution has been demonstrated and validated on a realistic simulated UAV platform in Gazebo simulations, where the UAV is operated at a high speed
Real-time Model Predictive Control for Aerial Manipulation
Full text linkThe rapid development in the field of Unmanned Aerial Vehicles (UAVs) is driven by new applications in agriculture, logistics, inspection and smart manufacturing. The future keys in these domains are the abilities to autonomously interact with the environment and with other robotic systems. This thesis is providing control engineering solutions to contribute to these key capabilities.
The first step of this thesis is to develop an understanding of the dynamic behavior of UAVs. For this purpose, dynamic and kinematic models are presented to describe a UAV's motion. This includes a kinematic model which is suitable for off-the-shelf UAVs and combines full 360° heading operation with a low computational complexity. The presented models are subsequently used to develop a nonlinear model predictive control NMPC strategy. In this context, the performance of several NMPC solvers and inequality constraint handling techniques is evaluated. The real-time capability and NMPC performance are validated with real AR.Drone 2.0 and DJI M100 quadrotors. This includes collision avoidance and advanced tracking scenarios. The design work-flow for the related control objectives and constraints is presented accordingly. As a next step, this UAV NMPC strategy is extended for a UAV with attached robotic arm. For this purpose, the forward kinematics of the robotic arm are developed and combined with the kinematic model of the UAV. The resulting NMPC strategy is validated in a grasping scenario with a real aerial manipulator. The final step of this thesis is the NMPC of cooperating UAVs. The computational complexity of such scenarios conflicts directly with the fast UAV dynamics. In addition, control objectives and system topologies can dynamically change. To address these challenges, this thesis presents the DENMPC software framework. DENMPC provides a computationally efficient central NMPC strategy that allows changing the control scenario at runtime. This is finally stated in the control of a real cooperative aerial manipulation scenario
Learning of Control Behaviours in Flying Manipulation
Full text linkMachine learning is an ever-expanding field of research with a wide range of potential applications. It has been increasingly used in different robotics tasks enhancing their autonomy and intelligent behaviour. This thesis presents how machine learning techniques can enhance the decision-making ability for control tasks in aerial robots as well as amplify the safety, thus broadly improving their autonomy levels.
The work starts with the development of a lightweight approach for identifying degradations of UAV hardware-related components, using traditional machine learning methods. By analysing the flight data stream from a UAV following a predefined
mission, it predicts the level of degradation of components at early stages. In that context, real-world experiments have been conducted, showing that such approach can be used as a safety system during different experiments, where the flight path of the vehicle is defined a priori. The next objective of this thesis is to design intelligent control policies for flying robots with highly nonlinear dynamics, operating in continuous state-action setting, using model-free reinforcement learning methods. To achieve this objective, first, the nuances and potentials of reinforcement learning have been investigated. As a result, numerous insights and strategies have been pointed out for crafting efficient reward functions that lead to successful learning performance. Finally, a learning-based controller is provided for controlling a hexacopter UAV with 6-DoF, to perform stable navigation and hovering actions by directly mapping observations to low-level motor commands. To increase the complexity of the given objective, the degrees of freedom of the robotic platform is upgraded to 7-DoF, using a flying manipulation as learning agent. In this case, the agent learns to perform a mission composed of take-off, navigation, hovering and end-effector positioning tasks. Later, to demonstrate the effectiveness of the proposed controller and its ability to handle higher number of degrees of freedom, the flying manipulation has been extended to a robotic platform with 8-DoF. To overcome several challenges of reinforcement learning, the RotorS Gym experimental framework has been developed, providing a safe and close to real simulated environment for training multirotor systems. To handle the increasingly growing complexity of learning tasks, the Cyber Gym Robotics platform has been designed, which extends the RotorS Gym framework by several core functionalities. For instance, it offers an additional mission controller that allows to decompose complex missions into several subtasks, thus accelerating and facilitating the learning process. Yet another advantage of the Cyber Gym Robotics platform is its modularity
which allows to seamlessly switch both, learning algorithms as well as agents. To validate these claims, real-world experiments have been conducted, demonstrating that the model trained in the simulation can be transferred onto a real physical robot
with only minor adaptations
A Combined Machine Learning Approach For the Engineering of Flexible Assembly Processes Using Collaborative Robots
Full text linkThe manufacturing industry is witnessing a rapid transformation from mass production to mass customization, driven by increasing consumer demand for highly personalized products. Collaborative robots (cobots) play a key role in enabling this shift, as their flexibility supports the dynamic and varied tasks necessary for producing high-mix, low-volume batches. However, traditional robot programming methods are not suited for unstructured environments with frequent task variations. These approaches quickly become cumbersome, prone to errors, and demand specialized robotic expertise. In response, Learning from Demonstration (LfD) has emerged as a promising paradigm for teaching tasks to robots by having them observe human demonstrations. This not only addresses the need for explicit programming but also allows non-experts to program robots efficiently. Nevertheless, practical
deployment of LfD in real industrial scenarios remains challenging, particularly when ensuring that customized robotics solutions still meet the high-performance standards associated with traditional mass production processes. This thesis addresses these challenges by (1) proposing a practical roadmap that guides both researchers and industry practitioners in transitioning from rigid, mass production–oriented robotic tasks to flexible, LfD-based mass customization workflows; (2) introducing a one-shot demonstration framework, DFL-TORO, which captures time-optimal and smooth trajectories from a single human demonstration; and (3) presenting a modular, standardized software framework integrating LfD methodologies in manufacturing systems. Through an in-depth case study and experimental validations, the thesis lays the foundation for bridging the gap between academic research in LfD and its real-world adoption in mass customization settings
Optinization of wireless energy transfer for mid-range distances
No full textThis thesis deals with the optimization of resonators for wireless power transfer by resonant magnetic coupling, where loosely coupled LC resonators are used for wireless power transmission. While systems that are currently available use discrete capacitors for frequency tuning, this work proposes self-resonating coils on a printed circuit board. For these coils, different three-dimensional (3D) Computer Aided Design (CAD) models are created and investigated using the method of finite elements. Finally, a set of inductively coupled and capacitively coupled self-resonant coils for the 6.78 MHz ISM band is found. Furthermore, the shielding of the coils with ferrite materials is discussed to maintain a high efficiency for any application. For both types, equivalent circuit models are derived and analyzed. Besides the coil optimization, a power amplifier is also proposed. During operation the optimal operational frequency is maintained by frequency tracking. The theoretical results are verified using a demonstrator circuit
Safety of Autonomous Cognitive-oriented Robots
No full textService robots shall very soon autonomously provide services in all spheres of life by executing demanding and complex tasks in dynamic, complex environments and by collaborating with human users. In order to push forward the understanding of the safety problem a novel classification of robot hazards is provided. The so-called object interaction hazards are derived which arise when environment objects interact with objects that are manipulated by a robot. Taking into account the current state-of-the-art, it can be stated that this denotes a novel problem area. However, it is already proposed the so-called dynamic risk assessment approach, which shall enable the robot to perceive the risk of current and upcoming situations. In order to realize such a risk-aware planning system for the first time, dynamic risk assessment is integrated within a cognitive architecture serving cognitive functions like anticipation, planning and learning. In this connection, action spaces (sets of possible upcoming situations) are dynamically anticipated assessed with regard to comprised risks. Though, (initial) knowledge about hazards is required in order to realize this. Therefore, a novel procedural model is developed for systematically generating a safety knowledge base. However, it can be assumed that the safety knowledge potentially lacks completeness. The application of AI-based approaches constitutes a noteworthy opportunity. For this reason, light is shed on strategically influential learning methods in safety-critical contexts.
Finally, this work describes the generation, integration, utilization, and maintenance of a system-internal safety knowledge base for dynamic risk assessment. It denotes an overall concept toward solving the advanced safety problem and confirms in principle the realization of a safe behavior of autonomous and intelligent systems.Sicherheit autonomer kognitivorientierter Roboter
Autonome mobile Serviceroboter sollen zukünftig selbstständig Dienstleistungen in allen Lebensbereichen erbringen, auch in direkter Nähe zum Menschen. Um das Verständnis für Sicherheit in der Robotik zu erwei-tern, wird zunächst eine neue Klassifizierung der möglichen Gefahren vorgenommen. Hiervon wird die Klasse der Objektinteraktionsgefahren abgeleitet. Diese Gefahren entstehen, wenn Objekte der Umgebung mit denen interagieren, die der Roboter greift und transportiert. In Anbetracht des aktuellen Standes der Sicherheits-technik in der Robotik wird klar, dass sich hier ein neues Problemfeld auftut. Grundsätzlich wurde bereits ein dynamischer Risikountersuchungsansatz vorgeschlagen, welcher den Roboter selbst befähigen soll, Situatio-nen hinsichtlich möglicher Gefahren zu untersuchen. Um dadurch eine risikobewusste Handlungsplanung erstmals zu realisieren, wird dieser in eine kognitive Architektur integriert, um kognitive Funktionen, wie Anti-zipation, Planen und Lernen zu nutzen. Hierbei werden mögliche Handlungsräume dynamisch antizipiert und mittels dynamischer Risikoanalyse auf mögliche Gefahren untersucht. Um (Objektinteraktions-) Gefahren mit Hilfe der dynamischer Risikountersuchung bestimmen zu können, bedarf es eines (initialen) Wissens über mögliche Gefahren. Aus diesem Grund wird ein Vorgehensmodell zur systematischen Erzeugung einer solchen Sicherheitswissensbasis entwickelt. Dieses Sicherheitswissen ist jedoch potentiell unvollständig. Daher stellt die Erweiterung und Verfeinerung desselben eine Notwendigkeit dar. Hierbei können die Ansätze aus dem Bereich der künstlichen Intelligenz als nützliche Möglichkeit wahrgenommen werden. Daher werden strate-gisch wichtige Lernmethoden hinsichtlich der Anwendung in einem sicherheitskritischen Kontext untersucht.
Die vorliegende Arbeit beschreibt die Erzeugung, die Integration, die Verwendung und die Aufrechterhaltung einer systeminternen Sicherheitswissensbasis zum Zwecke der dynamischen Risikountersuchung. Sie stellt hierbei ein Gesamtkonzept dar, dass zur Lösung des erweiterten Sicherheitsproblems beiträgt und somit die prinzipielle Realisierung des sicheren Betriebs von autonomen und intelligenten bestätigt
Design for meaning in products and services to foster eco-sufficient user behavior: exemplified by sharing goods
Full text link- …
