In this study, we demonstrate that an open-domain conversational system trained on idioms or figurative language generates more fitting responses to prompts containing idioms. Idioms are a part of everyday speech in many languages and across many cultures, but they pose a great challenge for many natural language processing (NLP) systems that involve tasks such as information retrieval (IR), machine translation (MT), and conversational artificial intelligence (AI). We utilized the Potential Idiomatic Expression (PIE)-English idiom corpus for the two tasks that we investigated: classification and conversation generation. We achieved a state-of-the-art (SoTA) result of a 98% macro F1 score on the classification task by using the SoTA T5 model. We experimented with three instances of the SoTA dialogue model—the Dialogue Generative Pre-trained Transformer (DialoGPT)—for conversation generation. Their performances were evaluated by using the automatic metric, perplexity, and a human evaluation. The results showed that the model trained on the idiom corpus generated more fitting responses to prompts containing idioms 71.9% of the time in comparison with a similar model that was not trained on the idiom corpus. We have contributed the model checkpoint/demo/code to the HuggingFace hub for public access.

This paper describes the system used by the Machine Learning Group of LTU in subtask 1 of the SemEval-2022 Task 4: Patronizing and Condescending Language (PCL) Detection. Our system consists of finetuning a pretrained text-to-text transfer transformer (T5) and innovatively reducing its out-of-class predictions. The main contributions of this paper are 1) the description of the implementation details of the T5 model we used, 2) analysis of the successes & struggles of the model in this task, and 3) ablation studies beyond the official submission to ascertain the relative importance of data split. Our model achieves an F1 score of 0.5452 on the official test set.

The main challenge of any mobile robot is to detect and avoid obstacles and potholes. This paper presents the development and implementation of a novel mobile robot. An Arduino Uno is used as the processing unit of the robot. A Sharp distance measurement sensor and Ultrasonic sensors are used for taking inputs from the environment. The robot trains a neural network based on a feedforward backpropagation algorithm to detect and avoid obstacles and potholes. For that purpose, we have used a truth table. Our experimental results show that our developed system can ideally detect and avoid obstacles and potholes and navigate environments.

Today, manufacturing companies are under competitive and rapid changes in the global market. It challenges companies working with material manufacturing and distribution networks at a global level to continually improve their business. The actors operating within a single supply chain should unite to increase their ability to compete and remain flexible, as a response to the effects of globalization. By creating business goals, sharing information, risks, and profits, all operators in the chain can benefit.

This case study has been carried out in one of Oxelösund Harbor's operations, namely SSAB's Steel Port, which they operate for SSAB in Oxelösund municipality. The steel harbor has an important role in SSAB's supply chain because it acts as SSAB's port of shipment where materials coming from SSAB in Borlänge and Oxelösund are stored until they are shipped to other ports in the world. Today, Oxelösund Harbor is experiencing inefficient information and material flow within the Steel Harbor's operations.

The purpose of the study is to investigate material handling in terms of which transport and storage methods are carried out in one of the steel ports operations, the coil terminal. Thereafter, the study will clarify what kind of waste there is and find improvements to streamline information and material flow when handling coils in the coil terminal.

The study was primarily conducted through observations, unstructured and to a lesser extent structured interviews, structured email conversations and data collection. The focus was on finding the possible areas that could become more efficient in practice and then finding the theory and literature that highlight the shortcomings found. Some theories that will be emphasized, because of their relevance, are Lean, inventory and supply chain theory.

The study's results and analysis show that the challenges that hinder the efficiency of the coil terminal are based on internal and external factors. Internal factors consist of unnecessary movement of materials, reworking in office and unused work shifts. External factors are caused by vendors in the form of information insufficiency and material input/output.

The results of the study conclude that the optimal solution is that the cooperation between the operators in the chain is improved so that flow of information becomes better and reduce uncertainty in delivery and shipping. Furthermore, the study encourages Oxelösund Harbor to collect data, in the form of statistics, concerning daily events in Stålhamnen, highlighting problems and costs that are hidden and start utilizing the unused work shift during the night, as well as reducing the stock.

This thesis demonstrates a technique for developing efficient applications interpreting spacial deep learning output using Hyper Dimensional Computing (HDC), also known as Vector Symbolic Architecture (VSA). As a part of the application demonstration, a novel preprocessing technique for motion using state machines and spacial semantic pointers will be explained. The application will be evaluated and run on a Google Coral edge TPU interpreting real time inference of a compressed object detection model.

In this article, a HUmanoid Robotic Leg (HURL) via the utilization of pneumatic muscle actuators (PMAs) is presented. PMAs are a pneumatic form of actuation possessing crucial attributes for the implementation of a design that mimics the motion characteristics of a human ankle. HURL acts as a feasibility study in the conceptual goal of developing a 10 degree-of-freedom (DoF) lower-limb humanoid for compliance and postural control, while serving as a knowledge basis for its future alternative use in prosthetic robotics. HURL’s design properties are described in detail, while its 2-DoF motion capabilities (dorsiflexion–plantar flexion, eversion–inversion) are experimentally evaluated via an advanced nonlinear PID-based control algorithm.

The presented work investigates the potential of utilizing commercially available Electric Ducted Fans (EDFs) as adhesion actuators, while providing a novel insight on the analysis of the adhesion nature related to negative pressure and thrust force generation against a target surface. To this goal, a novel EDF-based Vortex Actuation Setup (VAS) is proposed for monitoring important properties such as adhesion force, pressure distribution, current draw, motor temperature etc. during the VAS’ operation when placed in variable distances from a test surface. In addition, this work is contributing towards the novel evaluation of different design variables and modifications to original EDF structures, with the goal of analyzing their effect on the prototype VAS, while optimizing its adhesion efficiency for its future incorporation in a wall-climbing robot for inspection and repair purposes.

Robots are increasingly utilized in industry for their capability to perform repetitive,complex tasks in environments unsuitable for humans. This surge in robotic applicationshas spurred research into Multi-Robot Systems (MRS), which aim to tackle complex tasksrequiring collaboration among multiple robots, thereby boosting overall efficiency. However,MRS introduces multifaceted challenges that span various domains, including robot perception,localization, task assignment, communication, and control. This dissertation delves into theintricate aspects of task assignment and path planning within MRS.The first area of focus is on multi-robot navigation, specifically addressing the limitationsinherent in current Multi-Agent Path Finding (MAPF) models. Traditional MAPF solutionstend to oversimplify, treating robots as holonomic units on grid maps. While this approachis impractical in real-world settings where robots have distinct geometries and kinematicrestrictions, it is important to note that even in its simplified form, MAPF is categorized as anNP-hard problem. The complexity inherent in MAPF becomes even more pronounced whenextending these models to non-holonomic robots, underscoring the significant computationalchallenges involved. To address these challenges, this thesis introduces a novel MAPF solverdesigned for non-holonomic, heterogeneous robots. This solver integrates the hybrid A*algorithm, accommodating kinematic constraints, with a conflict-based search (CBS) for efficientconflict resolution. A depth-first search approach in the conflict tree is utilized to accelerate theidentification of viable solutions.The second research direction explores synergizing task assignment with path-finding inMRS. While there is substantial research in both decentralized and centralized task assignmentstrategies, integrating these with path-finding remains underexplored. This dissertation evaluatesdecoupled methods for sequentially resolving task assignment and MAPF challenges. Oneproposed method combines the Hungarian algorithm and a Traveling Salesman Problem (TSP)solver for swift, albeit suboptimal, task allocation. Subsequently, robot paths are generatedindependently, under the assumption of collision-free navigation. During actual navigation, aNonlinear Model Predictive Controller (NMPC) is deployed for dynamic collision avoidance. Analternative approach seeks optimal solutions by conceptualizing task assignment as a MultipleTraveling Salesman Problem (MTSP), solved using a simulated annealing algorithm. In tandem,CBS is iteratively applied to minimize the cumulative path costs of the robots.

In this paper, we study the multi-robot task assignment and path-finding problem (MRTAPF), where a number of robots are required to visit all given tasks while avoiding collisions with each other. We propose a novel two-layer algorithm SA-reCBS that cascades the simulated annealing algorithm and conflict-based search to solve this problem. Compared to other approaches in the field of MRTAPF, the advantage of SA-reCBS is that without requiring a pre-bundle of tasks to groups with the same number of groups as the number of robots, it enables a part of robots needed to visit all tasks in collision-free paths. We test the algorithm in various simulation instances and compare it with state-of-the-art algorithms. The result shows that SA-reCBS has a better performance with a higher success rate, less computational time, and better objective values.

This paper presents a complete system architecture for multi-robot coordination for unbalanced task assignments, where a number of robots are supposed to visit and accomplish missions at different locations. The proposed method first clusters tasks into clusters according to the number of robots, then the assignment is done in the form of one-cluster-to-one-robot, followed by solving the traveling salesman problem (TSP) to determine the visiting order of tasks within each cluster. A nonlinear model predictive controller (NMPC) is designed for robots to navigate to their assigned tasks while avoiding colliding with other robots. Several simulations are conducted to evaluate the feasibility of the proposed architecture. Video examples of the simulations can be viewed at https://youtu.be/5C7zTnv2sfo and https://youtu.be/-JtSg5V2fTI?si=7PfzZbleOOsRdzRd. Besides, we compare the cluster-based assignment with a simulated annealing (SA) algorithm, one of the typical solutions for the multiple traveling salesman problem (mTSP), and the result reveals that with a similar optimization effect, the cluster-based assignment demonstrates a notable reduction in computation time. This efficiency becomes increasingly pronounced as the task-to-agent ratio grows.

In this letter, a floating robotic emulation platform is presented with an autonomous maneuverability for a virtual demonstration of a satellite motion. Such a robotic platform design is characterized by its friction-less, levitating, yet planar motion over a hyper-smooth surface. The design of the robotic platform, integrated with the sensor and actuator units, is briefly described, including the related component specification along with the mathematical model, describing its dynamic motion. Additionally, the article establishes a nonlinear optimal control architecture consisting of a unified model predictive approach for the overall manoeuvre tracking. The efficacy of the proposed modeling and control scheme is demonstrated in multiple experimental studies, where it is depicted that the proposed controller has the potential to address a precise point-to-point manoeuvre with terminal objectives, as well as an excellent path following capability. The proposed design is validated with extensive experimental studies, and it is supported with related results.

The paper introduces a novel framework for safe and autonomous aerial physical interaction in industrial settings. It comprises two main components: a neural network-based target detection system enhanced with edge computing for reduced onboard computational load, and a control barrier function (CBF)-based controller for safe and precise maneuvering. The target detection system is trained on a dataset under challenging visual conditions and evaluated for accuracy across various unseen data with changing lighting conditions. Depth features are utilized for target pose estimation, with the entire detection framework offloaded into low-latency edge computing. The CBF-based controller enables the UAV to converge safely to the target for precise contact. Simulated evaluations of both the controller and target detection are presented, alongside an analysis of real-world detection performance.

In December 2015 the Paris Agreement was signed in order to create a better global climate. According to the agreement, all countries 'emissions shall be reported to the UN. To be able to do this current emission levels must be known, which means measurement and calculation of emissions are necessary.

The state-owned mining company Luossavaara Kiirunavaara AB (LKAB) operates iron ore mines in Northen Sweden. The iron ore products produced at the KK4 coal mill in Kiruna, consists of some additives that form carbon dioxide when burned. The amount of additives burned is measured using two flowmeters. To ensure that these meters have approved accuracy, annual calibrations are performed where measured volume is compared to a reference volume. In order to ensure that LKAB's calibration method works, another type of calibration was performed by an external company. The results between the two calibration methods differ significantly, which led to this report.

The purpose of this work is to develop improvement proposals for LKAB's calibration method and to present suggestions on how LKAB can create better conditions for the other so called transit time calibration method. By analyzing each method and performing measurements, tests and calculations, weaknesses are noted. Tests show that a weakness of LKAB's calibration is that the level measurement method used is person-dependent. Level measurement with laser range gauge is tested with good results, why this type of measurement is recommended in the future. The main weakness with transit time method is primarily that it is sensitive to variation of the inner diameter of the pipe. In order for this method to provide reliable results, rebuilding of pipes and carefully measured internal diameter are required. Accurate flow measurement means that current carbon dioxide emissions can be calculated, which enables continued work towards set environmental goals. Flowmeters also have a significant role in product quality in KK4, which means that this report can contribute to higher quality and economic profitability.

Earth-moving machines, such as wheel loaders, are a type of heavy-duty machinery used within the construction industry to perform vital tasks, such as digging, transporting, and mining applications. One of these tasks is the short-loading cycle, where an operator manoeuvres the wheel loader to move material from a pile to the tipping body of a dump truck, through navigation, scooping, and dumping. The short-loading cycle is a repetitive task performed in high quantities, often as part of a larger refinement process, making it interesting for automation.

The main objective of this thesis work is to investigate challenges facing the automation of the short-loading cycle, focusing in particular on subtasks that can be efficiently addressed with deep learning methods. A secondary objective is to examine how alternative development paths, such as scale models, or simulations, can be used to enable data-driven automation of the short-loading cycle, as directly experimenting on real vehicles has a high associated cost when large numbers of timesteps are needed to gather enough data.

To investigate the two objectives, the literature is systematically reviewed to identify research gaps, challenges, and the usage of deep learning techniques. Secondly, a set of deep learning techniques is investigated to address perception and actuation problems identified as challenging and important for the automation of the short-loading cycle.

The investigation of deep learning techniques involves training and validating a realtime object detector neural network to identify key components (wheels, tipping body, and cab) on a scale model dump truck while testing on a real vehicle. This resulted in a localisation and classification degradation of only 14% between the scale model and the real dump truck, with no additional training. In addition, an examination to minimize the annotation workload of humans found that it is possible to decrease the workload by 95% while still retaining similar detection performance by leveraging linear interpolation.

Lastly, this thesis presents an investigation regarding the usage of reinforcement learning for navigation during the short-loading cycle. The results indicate that training the agent in simulation is currently required as the agent obtains the maximum reward after timesteps in the order of millions before being capable of performing the task. The results suggest that the trained agent is capable of bridging the gap between simulation and reality to complete a simplified version of the navigation task during the short-loading cycle.

The experiments presented in this thesis provide proof of concept that indicates deep learning techniques can aid in the realisation of an autonomous solution. Moreover, the results show that development paths allowing for experiments providing large numbers of timesteps can facilitate the practical use of such techniques.

The short-loading cycle is a construction task where a wheel loader scoops material from a nearby pile in order to move that material to the tipping body of a dump truck. The short-loading cycle is a vital task performed in high quantities and is often part of a more extensive never-ending process to move material for further refinement. This, together with the highly repetitive nature of the short-loading cycle, makes it a suitable candidate for automation. However, the short-loading cycle is a complex task where the mechanics of the wheel loader together with the interaction between the wheel loader and the environment needs to be considered. This must be achieved while maintaining some productivity goal and, concurrently, minimizing the used energy. The main objective of this work is to analyze the short-loading cycle, assess the current state of research in this field, and discuss the steps required to progress towards a minimal viable product consisting of individual automation solutions that can perform the short-loading cycle well enough to be used by early adopters. This is achieved through a comprehensive literature study and consequent analysis of the review results. From this analysis, the requirements of an MVP are defined and some gaps which are currently hindering the realization of the MVP are presented.

The short-loading cycle is a repetitive task performed in high quantities, making it a great alternative for automation. In the short-loading cycle, an expert operator navigates towards a pile, fills the bucket with material, navigates to a dump truck, and dumps the material into the tipping body. The operator has to balance the productivity goal while minimising the fuel usage, to maximise the overall efficiency of the cycle. In addition, difficult interactions, such as the tyre-to-surface interaction further complicate the cycle. These types of hard-to-model interactions that can be difficult to address with rule-based systems, together with the efficiency requirements, motivate us to examine the potential of data-driven approaches. In this paper, the possibility of teaching an agent through reinforcement learning to approach a dump truck's tipping body and get in position to dump material in the tipping body is examined. The agent is trained in a 3D simulated environment to perform a simplified navigation task. The trained agent is directly transferred to a real vehicle, to perform the same task, with no additional training. The results indicate that the agent can successfully learn to navigate towards the dump truck with a limited amount of control signals in simulation and when transferred to a real vehicle, exhibits the correct behaviour. 

A new challenging area of research in autonomous systems focuses on the collaborative multi-agent exploration of unknown environments where a reliable communication infrastructure among the robotic platforms is absent. Factors like the proximity between agents, the characteristics of the network nodes, and environmental conditions can significantly impact data transmission in real-world applications. We present a novel decentralized collaborative architecture based on multi-agent reinforcement learning to address this challenge. In this framework, homogeneous agents autonomously decide to communicate or not, that is whether to share locally collected maps with other agents in the same communication networks or to navigate and explore the environment further. The agents' policies are trained using the heterogeneous-agent proximal policy optimization (HAPPO) algorithm and through a novel reward function that balances inter-agent communication and exploratory behaviors. The proposed architecture enhances mapping efficiency and robustness while minimizing inter-agent redundant data transmission. Finally, this paper demonstrates the advantages of the investigated approach compared to a strategy that does not incentivize communicative behaviors.

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.

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.

Negotiations of different kinds are used to trade goods and services. Within these, the creation of a signed agreement or contract that is binding for the agreeing parties helps also the gathering of evidence that can be used in case of disputes and for adjudication. Traditionally, contracts are established on paper agreements that are signed by all the involved parties and by a law enforcement entity that ensure its legality in a court of law. These contracts have evolved with the introduction of Information Technology (IT) where the negotiation of goods and services is mainly virtual and/or automatized. The consistency and processing time of the computers allow for negotiations to be more efficient than ever. 

Digitalized negotiations allow for auctioning systems providing a mechanism to efficiently match demand and supply in the exchange of goods and services. Such suctioning systems allow multiple users to iteratively or non-iteratively compete against one another to achieve allocative efficiency. Lately, digitalized auctions are implemented using Blockchain systems with the use of Smart Contracts to archieve decentralization. These are implemented as a digital script that may encode any set of rules written as code, with the validity of the code being enforced by the Blockchain's consensus mechanism. These Smart Contracts computations however tend to be expensive when executed and limited by the blocksize.  

This thesis studies the creation of digitized negotiation protocols and contract definition following the needs of traditional trading and auctioning systems. We investigate the use of Ricardian Contracts for flexible representation of rights and obligations of entities in the context of circular economy in both single and multi-attribute auctions. We analyze the implication of digitized agreements in the context of data sharing. Furthermore, we analyze how usage control policies can be represented into Ricardian Contracts in the context of intellectual property protection, compliance with regulations, and digital rights management.Finally, we analyze the properties that a system that supports the mentioned models should have and how to implement it in the context of distributed auctioning systems by contrasting available state-of-the-art. 

The main contributions of the thesis are: (1) The creation of a multi-attribute auctioning protocol for the circular economy which implements Ricardian Contracts for the representation of rights and obligations. (2) A method to negotiate obligations and access provisions with multi-level Ricardian contracts, and automatically enforce those provisions with access control. (3) A state-of-art analysis on distributed and decentralized auctioning systems where the key properties of auctioning systems are identified and are evaluated against the current implementations.

In this paper, we define a multi-attribute auctioning system for the circular economy and the trade of products, components and materials subject to recycling. The increasing popularity of auctioning systems for buying and selling goods has led to the adaptation of them to diverse and particular scenarios, many of which require support for attributes like delivery time, quality, etc. Such attributes allow for more explicit and precise negotiations than traditional auctioning systems where only price is taken into account. The circular economy concept replaces end-of-life with the reuse of various goods, aiming to keep as much value as possible of any asset. By allowing users to adjust attributes in multi-step negotiations according to their economic and ecological needs, better deals can be achieved. We address this potential with our multi-attribute, and multi-step auctioning system. The system is based on transparency and fairness principles, and addresses requirements for flexibility in what attributes can be used, and the need for a semi-transparent auctioning procedure. We present a winner determination approach based on scoring protocol based on weights for different input attributes. Our auctioning system uses a signature chain data structure to provide transaction traceability. We demonstrate using a generic example that the proposed system supports simple and flexible multi-attribute auctions.

In this paper, we define an auction protocol and implementation of a decentralized and distributed auctioning platform for demand-supply matching of components and materials subject to recycling. Auctioning is the process of buying and selling goods or services by offering them up for bid, taking bids, and then selling the item to a winner according to preset rules. Some auction types relies on sealed-bids while others implement an open-bid procedure that allows for multiple bids before a known or unknown deadline. The English auction and its variants is most common type. It relies on that the current highest bid is always available to potential bidders. Online auction platforms are nowadays used to trade various services and goods. We address the iterative negotiations in form of auction bids aiming at signed contractual agreements stated in legal prose and captured by Ricardian contracts. We evaluate our model towards requirements including privacy, transparency and fairness in terms of acknowledged publishing of highest bids and the ordered arrival of individual bids. In addition, we demonstrate the mechanisms for distributed storage of cryptographically signed Ricardian contracts, omitting the need for trusting the auctioneer or relying on a trusted third party for this storage.

Data is important for the industry to take advantage of digitalization, realize automation, assure quality, and more. Values from data are not only created individually by companies, but also in eco-systems in which data is shared among participating organizations. Secure data sharing is essential in such eco-systems to prevent unauthorized access and use of the data. Usage control extends traditional access control with restrictions concerned with requirements that pertain to data processing contractual obligations, rather than data access provisions only. Thus, usage control is relevant in the context of intellectual property protection, compliance with regulations, and digital rights management. This paper presents a method to negotiate contractual obligations and access provisions, and automatically enforce those provisions with access control. Finalized negotiations establish Ricardian contracts at two levels; a superordinate level with a connected subordinate level. These contracts contain provisions in terms of access control attributes. Using our implementation of a negotiation engine we demonstrate the automatic creation of NIST Next Generation Access Control (NGAC) access control policies. Our negotiation engine uses a lightweight model for the storage of an unforgeable and immutable log of the established contracts based on digital signatures and hashing.

The possibility of using multiple autonomous agents for completing various tasks has gathered a lot of interest in the last years. Using multiple agents does however come with the problems of choosing how to allocate the tasks and how the intelligence of the individual agents should be designed. 

In this thesis a market inspired auction system for task allocation and an approach for generating a behavior tree creating the autonomous behavior using a library of available behaviors will be established. The validity of the proposed approach is demonstrated in simulation and with multiple experiments in a laboratory environment. The result does show that the approach promisingly solves the stated problems.

This thesis introduces a novel framework for enabling intelligent machines and robots with the fifth-generation (5G) cellular network technology. Autonomous robots, such as Unmanned Aerial Vehicles (UAVs), Autonomous Guided Vehicles (AGVs), and more, can notably benefit from multi-agent collaboration, human supervision, or operation guidance, as well as from external computational units such as cloud edge servers, in all of which a framework to utilize reliable communication infrastructure is needed. Autonomous robots are often employed to alleviate humans by operating demanding missions such as inspection and data collection in harsh environments or time-critical operations in industrial environments - to name a few. For delivering data to other robots to maximize the effectiveness of the considered mission, for executing complex algorithms by offloading them into the edge cloud, or for including a human operator/supervisor into the loop, the 5G network and its advanced Quality of Service (QoS) features can be employed to facilitate the establishment of such a framework. This work focuses on establishing a baseline for integrating various time-critical robotics platforms and applications with a 5G network. These applications include offloading computationally intensive Model Predictive Control (MPC) algorithms for trajectory tracking of UAVs into the edge cloud, adapting data sharing in multi-robot systems based on network conditions, and enhancing network-aware surrounding autonomy components. We have identified a set of key performance indicators (KPIs) crucially affecting the performance of network-dependent robots and applications. We have proposed novel solutions and mechanisms to meet these requirements, which aim to combine traditional robotics techniques to enhance mission reliability with the exploitation of 5G features such as the QoS framework. Ultimately, our goal was to develop solutions that adhere to the essential paradigm of co-designing robotics with networks. We thoroughly evaluated all presented research using real-life platforms and 5G networks.

Offloading processes responsible for a robot’s control operation to external computational resources has been in the spotlight for many years. The vision of having access to a full cloud cluster for any autonomous robot has fueled many scientific fields. Such implementations rely strongly on a robust communication link between the robot and the cloud and have been tested over numerous network architectures. However, various limitations have been highlighted upon the realization of such platforms. For small-scale local deployments, technologies such as Wi-Fi, Zigbee, and blacktooth are inexpensive and easy to use but suffer from low transmit power and outdoor coverage limitations. In this study, the offloading time-critical control operations for an unmanned aerial vehicle (UAV) using cellular network technologies were evaluated and demonstrated experimentally, focusing on the 5G technology. The control process was hosted on an edge server that served as a ground control station (GCS). The server performs all the computations required for the autonomous operation of the UAV and sends the action commands back to the UAV over the 5G interface. This research focuses on analyzing the low-latency needs of a closed-loop control system that is put to the test on a real 5G network. Furthermore, practical limitations, integration challenges, the intended cellular architecture, and the corresponding Key Performance Indicators (KPIs) that correlate to the real-life behavior of the UAV are rigorously studied.

Over the last decade, the integration of robots into various applications has seen significant advancements fueled by Machine Learning (ML) algorithms, particularly in autonomous and independent operations. While robots have become increasingly proficient in various tasks, object instance recognition, a fundamental component of real-world robotic interactions, has witnessed remarkable improvements in accuracy and robustness. Nevertheless, most existing approaches heavily rely on prior information, limiting their adaptability in unfamiliar environments. To address this constraint, this thesis introduces the Segment and Learn Semantics (SaLS) framework, which combines video object segmentation with Continual Learning (CL) methods to enable semantic understanding in robotic applications. The research focuses on the potential application of SaLS in mobile robotics, with specific emphasis on the TORO robot developed at the Deutsches Zentrum für Luft- und Raumfahrt (DLR). Evaluation of the proposed method is conducted using a diverse dataset comprising various terrains and objects encountered by the TORO robot during its walking sessions. The results demonstrate the effectiveness of SaLS in classifying both known and previously unseen objects, achieving an average accuracy of 78.86% and 70.78% in the CL experiments. When running the whole method in the image sequences collected with TORO, the accuracy scores were of 75.54% and 84.75%, for known and unknown objects respectively. Notably, SaLS exhibited resilience against catastrophic forgetting, with only minor accuracy decreases observed in specific cases. Computational resource usage was also explored, indicating that the method is feasible for practical mobile robotic systems, with GPU memory usage being a potential limiting factor. In conclusion, the SaLS framework represents a significant step forward in enabling robots to autonomously understand and interact with their surroundings. This research contributes to the ongoing development of robotic systems that can operate effectively in unstructured environments, paving the way for more versatile and capable autonomous robots.

The methods used today for the commissioning and validation of industrial machines requires theconstruction of physical prototypes. Those prototypes help the engineers to e.g. validate if theprogram code meant to control a machine works as intended. In recent years the development ofnew techniques for the commissioning and validation of industrial machines has changed rapidlythanks to the development of new software. The method used in this thesis is called simulationin the loop. Another method that can be benecial to use is hardware in the loop. Using thosemethods for the commissioning of a machine is called virtual commissioning. The simulation inthe loop method is used to simulate both the machine and the control system that operate thatmachine. This is called a digital twin, a virtual copy of the physical hardware and its control systemthat can be used without the need for a real prototype to be available.The software used in this thesis comes all from the company Siemens and those are TIA Portal,Mechatronics Concept Designer, SIMIT and PLCSim Advanced. By using those programs it waspossible to build a digital twin with rigid body dynamics and its control system of the industrialmodel that was given by the company Renholmen AB. This model contained all the necessarycomponents needed for a virtual commissioning project to be done without the need to be at thefactory oor.The results showed that it was possible to achieve a real time simulation, allowing the possibilityto trim the controller parameters without the need of a physical prototype. Design errors were alsofound thanks to the results of the simulation.This new technique has shown to be a useful tool due to most of the work could be done on a digitalmodel of the machine. Simulations can reduce the time to market for industrial machines and alsohelp engineers to validate and optimize the product at an early stage. This tool that can be usedto validate industrial machines before they are created.

This thesis treats the development of a control system for an inverse soil conditioner prototype. A simulation model was created to develop a control system with the purpose of validation and verification of the prototypes efficacy. The simulation model is created in Simulink, where a part of the soil conditioner is imported as a solid model, which then is coupled to a model of a hydraulic system. In the simulation a control system and regulator were implemented and tuned. when the software was test-ready the hardware-interface was tested to validate that the current software could receive inputs and send meaningful outputs, and then real movements were logged to validate the software function for the machine.

The results of this project can then be summarized as a simulation model, a control system, and a solid basis for real world verification are completed.

In this paper, the discharge voltage is utilized as a critical indicator towards the probabilistic estimation of the Remaining Useful Life until the End-of-Discharge of the Lithium-Polymer batteries of unmanned aerial vehicles. Several discharge voltage histories obtained during actual flights constitute the in-house developed training dataset. Three data-driven prognostic methodologies are presented based on state-of-the-art as well as innovative mathematical models i.e. Gradient Boosted Trees, Bayesian Neural Networks and Non-Homogeneous Hidden Semi Markov Models. The training and testing process of all models is described in detail. Remaining Useful Life prognostics in unseen data are obtained from all three methodologies. Beyond the mean estimates, the uncertainty associated with the point predictions is quantified and upper/lower confidence bounds are also provided. The Remaining Useful Life prognostics during six random flights starting from fully charged batteries are presented, discussed and the pros and cons of each methodology are highlighted. Several special metrics are utilized to assess the performance of the prognostic algorithms and conclusions are drawn regarding their prognostic capabilities and potential.

Many technologies have been used in education. Such technologies fall under three main types: learning management systems; education data mining; and AI-enabled technologies. This report focuses on the use of robotics in interactive education. Over the past few years, interest in utilization of robotics in education has increased. Multiple attempts have been made, globally, in order to introduce robotics in education. Our report reveals that robotics have been used in education either in front scene acting as a teacher or in back scene supporting the teaching process. Our report also reveals that robotics are able to address unsolved educational issues such as achievement gaps and teachers gaps, in addition to the assistance it provides in some specific use cases. Further research efforts are indeed required to fully understand the exact role, current and future, or robotics in education. The report also introduce some challenges in using robotics in education e.g., Communication breakdowns, navigation capabilities, and the feeling of remote students

This thesis analyses the background of the kinematics and control of an articulated robotic arm in order to control it’s motor controllers with a custom controller using PLC programming.

The goal was to create a MATLAB simulation of a manipulator and then establish a working configuration. This would then later be used by the division of Signals and Systems at LTU to educate students in PLC programming for autonomous setups with the robotic manipulator.

The thesis was successful following the schedule established at the beginning and most of the objectives were accomplished. The results were a functioning control framework of Siemens PLC that could control the manipulators motor controllers to preform pickand place tasks in conjunction with a conveyor belt.

As robots are increasingly integrated into large and dynamic environments alongside humans, there is a pressing need for efficient onboard solutions to fundamental robotic operations, such as navigation and decision-making. Existing solutions often rely on computationally intensive processes that do not scale well in larger environments, leading to long computation times. This can result in unsafe and non-adaptive behaviors, as during the planning phase the robot continues to move along an old and potentially dangerous path increasing the risks of accidents or emergency.This thesis addresses this challenge by developing human-inspired light-weight methods, that enhance robotic navigation and environment understanding.

The central framework presented in this thesis introduces a novel human-like method for navigation and environment segmentation using 2D grid maps, focusing on extracting structural-semantics, such as intersections, pathways, dead ends, and paths to unexplored areas. The framework also generates sparse topometric maps for lightweight robotic navigation by using structural-semantic information. Compared to the state-of-the-art, where map segmentation either utilizes features that are specific to some indoor environments or segments into arbitrary regions that do not convey semantically meaningful information about the environment, the semantic topometric map captures structural-semantic information, which can easily be utilized by robots in a variety of missions. The proposed framework has been validated on multiple maps of different sizes and types of environments. In comparison with the state-of-the-art topological maps generated by Voronoi-based solutions, the proposed framework shows a significant reduction in complexity and computation times required in solving navigation problems. 

The utility of structural semantics is demonstrated through a novel autonomous exploration strategy that integrates structural-semantic information with conventional metric data for goal/frontier selection and employs the semantic topometric map for navigating to a frontier. The effectiveness of the exploration strategy is demonstrated in real-world experiments, showcasing improved exploration speed and computational efficiency compared to frontier-based exploration methods using only metric information. 

In order to enable the methods presented in this thesis to operate over 3D maps, this thesis introduces an approach for converting 3D voxel maps into 2D occupancy maps augmented with height and slope information. Moreover, a method for converting paths generated in 2D into 3D paths is proposed. This allows for the use of structural-semantic segmentation and efficient topometric map-based navigation planning for both UAVs and UGVs. These contributions together enable lightweight and fast environment segmentation and navigation planning for a multitude of robot types, and leveraging structural-semantic information leads to a more human-like approach toward robotic navigation and environment understanding. 

This article introduces a novel approach to constructing a topometric map that allows for efficient navigation and decision-making in mobile robotics applications. The method generates the topometric map from a 2D grid-based map. The topometric map segments areas of the input map into different structural-semantic classes: intersections, pathways, dead ends, and pathways leading to unexplored areas. This method is grounded in a new technique for intersection detection that identifies the area and the openings of intersections in a semantically meaningful way. The framework introduces two levels of pre-filtering with minimal computational cost to eliminate small openings and objects from the map which are unimportant in the context of high-level map segmentation and decision making. The topological map generated by GRID-FAST enables fast navigation in large-scale environments, and the structural semantics can aid in mission planning, autonomous exploration, and human-to-robot cooperation. The efficacy of the proposed method is demonstrated through validation on real maps gathered from robotic experiments: 1) a structured indoor environment, 2) an unstructured cave-like subterranean environment, and 3) a large-scale outdoor environment, which comprises pathways, buildings, and scattered objects. Additionally, the proposed framework has been compared with state-of-the-art topological mapping solutions and is able to produce a topometric and topological map with up to 92% fewer nodes than the next best solution. The method proposed in this article has been implemented in the robotics framework ROS and is open-sourced. The code is available at: https://github.com/LTU-RAI/GRID-FAST.

In this article, we introduce a novel strategy for robotic exploration in unknown environments using a semantic topometric map. As it will be presented, the semantic topometric map is generated by segmenting the grid map of the currently explored parts of the environment into regions, such as intersections, pathways, dead-ends, and unexplored frontiers, which constitute the structural semantics of an environment. The proposed exploration strategy leverages metric information of the frontier, such as distance and angle to the frontier, similar to existing frameworks, with the key difference being the additional utilization of structural semantic information, such as properties of the intersections leading to frontiers. The algorithm for generating semantic topometric mapping utilized by the proposed method is lightweight, resulting in the method’s online execution being both rapid and computationally efficient. Moreover, the proposed framework can be applied to both structured and unstructured indoor and outdoor environments, which enhances the versatility of the proposed exploration algorithm. We validate our exploration strategy and demonstrate the utility of structural semantics in exploration in two complex indoor environments by utilizing a Turtlebot3 as the robotic agent. Compared to traditional frontier-based methods, our findings indicate that the proposed approach leads to faster exploration and requires less computation time.

This article introduces a novel method for converting 3D voxel maps, commonly utilized by robots for localization and navigation, into 2D occupancy maps for both autonomous aerial vehicles (AAVs) and autonomous ground vehicles (AGVs). The generated 2D maps can be used for more efficient global navigation for both AAVs and AGVs, in enabling algorithms developed for 2D maps to be useful in 3D applications, and allowing for faster transfer of maps between multiple agents in bandwidth-limited scenarios. During the 3D to 2D map conversion, the method conducts safety checks with respect to the robot's safety margins. This ensures that an aerial or ground robot can navigate safely, relying primarily on the 2D map generated by the method. Additionally, the method extracts the height of navigable free space and a local estimate of the slope of the floor from the 3D voxel map. The height data is utilized in converting paths generated using the 2D map into paths in 3D space for both AAVs and AGVs. The slope data identifies areas too steep for a ground robot to traverse, marking them as occupied, thus enabling a more accurate representation of the terrain for ground robots. The proposed method is compared to the existing state-of-the-art fixed projection method in two different environments, over static maps and with progressively expanding maps. The methods proposed in this article have been implemented in the widely-used robotics frameworks ROS and ROS2, and are open-sourced.

This paper reviews the application of AI in maintenance and inspections. It gives an overview of the development of AVs and distant inspection operations for industrial assets using unmanned aerial vehicles (UAVs). It discusses the use of AVs in infrastructure inspection and explain the types of sensors used for these applications. It explains how autonomous robots, including drones, are currently used in various industrial settings for inspection and maintenance. The paper concludes by discussing the use of AI in predictive maintenance.

Free-flying robots are considered a valuable and emerging tool to support astronauts in their daily tasks in space facilities. This work presents the design and development of a free-flying robot as well as a self-contained mechanism that allows its docking for storage and tank refuelling. More specifically, this study presents a floating robotic emulation platform for a simulated demonstration of satellite mobility in orbit. Friction-less, levitating, yet flat motion across a hyper-smooth surface characterizes the robotic platform design. Moreover, the docking mechanism has been designed and developed for the free-flying robot to automate the docking and refuelling processes. The mechanism is divided into two main components, one fixed and one placed on the robot, where the major merit of the proposed system is that it addresses both the tank connection subsystem for the refuelling as well as the subsystem for the dock and repel phases. The former is enabled through the use of an actuated coupling support structure between the air tank and the external outlet, while the latter is enabled with the use of an electromagnetic connection support structure. Finally, preliminary hardware developments have been performed for the proposed robotic systems, demonstrating it's usefulness and effectiveness.

This paper presents an exploratory approach of the fetal heart rate (FHR) analysis, aiming to highlight potential limitations of the current predictive modeling attempts. To do so, a set of features that are usually encountered in FHR analysis as well as features extracted using a variant of symbolic aggregate approximation were projected onto a lower-dimensional space where patterns can easily be discerned. The results show, both in a qualitative and a quantitative manner, that there is high overlap between the classes that are formed using solely the umbilical cord pH information, irrespective of the selected dimensionality reduction method. These findings suggest that there is probably a limit to the performance expectation of the current pH-based systems and that alternative approaches should be also pursued to enhance the utility of computer-based decision support technologies.

Graden av automation inom industrin ökar ständigt, inom fordonstillverkning fasas allt fler produktionsnära personal ut till förmån för automatiska system och robotar. På Scanias hyttfabrik i Oskarshamn är stora delar av produktionen automatiserad och det söks ständigt efter nya möjligheter till effektivisering via automation. Detta examensarbete syftade till att utvärdera om kvalitetsinspektionen efter applicering av grundfärg kan automatiseras. Genom att analysera processen genomfördes utvärderingar av två automatiska inspektionssystem som i dag används inom fordonstillverkning.

Historisk data analyserades för att identifiera vad ett potentiellt automatiskt inspektionssystem bör klara av. I samband med detta samlades information in för att identifiera vad dessa inspektionssystem faktiskt klarar av. Detta ställdes emot varandra och föranledde en investeringskalkyl för att ta fram en återbetalningstid, samt för att upptäcka vilka monetära besparingar det går att erhålla via ett automatiskt inspektionssystem.

Analys genomfördes där fördelar och nackdelar med både dagens arbetssätt samt ett eventuellt framtida arbetssätt med automatiska system vägdes samman. Dessutom analyserades det till vilken grad kvalitetinspektionsprocessen hos Scania kan automatiseras, och vilka kvalitativa vinningar det leder till. I tillägg återger examensarbetet generella slutsatser på hur automatiska inspektionssystem fungerar samt presenterar resultat på om tekniken i dagsläget är mogen, och till vilken utsträckning.

Avslutningsvis presenteras rekommendationer till Scania utifrån möjligheter och potentiella besparingar. Rekommendationen är att inte investera i ett automatiskt inspektionssystem, eftersom examensarbetet kommit fram till att inget av de utvärderade systemen kan operera på en tillfredsställande nivå. Tekniken är det vill säga inte ännu mogen för att automatisera denna typ av process.

During the last decade, the efforts in space exploration have increased massively and led to a need for new ways to examine planets and other celestial bodies. The modern tendency is to create spacecraft able to scout the surface from a higher point of view, where drones have shown to be most helpful. Even if the benefits brought by this type of technology are considerable, the challenges are still difficult to overcome. This article presents a comprehensive literature review on drone technologies for planetary exploration, focusing mainly on the difficulties encountered. Considerable complications derive from the unknown environment, affecting most of the design, the mathematical model of the body, its controllability, and overall levels of autonomy. Various solutions to these challenges are proposed based on past and future missions. Furthermore, a look into the future gives an idea of possible technological developments and ways to provide the most efficient aerial exploration of other planets.

State and specifically location estimation is a core concept in automation and is a well-researched field. One such estimation technique is Moving Horizon Estimation (MHE). In this thesis, the MHE variant single-shooting estimation will estimate the location and velocity of a moving object. The moving object is equipped with an  Inertial Measuring Unit (IMU)  measuring acceleration and angular velocity. This thesis will explore pre-integrating the IMU measurement on the device attached to the moving object and using them in another device running the MHE. The acceleration and angular velocity measurements are measured in the local frame of reference of the moving object, rotating the measurement to a global frame of reference requires a known rotation of the tracked object, finding this rotation is also a task in this thesis. This thesis found the presented theory ill-equipped to estimate the object's state without an angle measurement, this thesis assumed any such measurement is made from a magnetometer but the solution presented is not biased towards any other method of measuring angles. With the addition of an angel measurement, the estimation performs at a decimeter-level precision for location.

In this work, we develop the transfer learning (TL) of reinforcement learning (RL) for the robotic skill of throwing a ball into a basket, from a computer simulated environment to a real-world implementation. Whereas learning of the same skill has been previously explored by using a Programming by Demonstration approach directly on the real-world robot, for our work, the model-based RL algorithm PILCO was employed as an alternative as it provides the robot with no previous knowledge or hints, i.e. the robot begins learning from a tabula rasa state, PILCO learns directly on the simulated environment, and as part of its procedure, PILCO models the dynamics of the inflatable, plastic ball used to perform the task. The robotic skill is represented as a Markov Decision Process, the robotic arm is a Kuka LWR4+, RL is enabled by PILCO, and TL is achieved through policy adjustments. Two learned policies were transferred, and although the results show that no exhaustive policy adjustments are required, large gaps remain between the simulated and the real environment in terms of the ball and robot dynamics. The contributions of this thesis include: a novel TL of RL framework for teaching the basketball skill to the Kuka robotic arm; the development of a pythonised version of PILCO; robust and extendable ROS packages for policy learning and adjustment in a simulated or real robot; a tracking-vision package with a Kinect camera; and an Orocos package for a position controller in the robotic arm.

In this article, we address the task of developing a unified solution that incorporates quadruped and aerial robots for planetary exploration missions. The designing process takes recommendations provided by Boston Dynamics for building custom payloads for the Spot robot, as well as its kinematic constraints. The unification task itself encompasses design of a passive drone landing platform as a hardware link between the Spot robot and the drone, which has active locking and unlocking capabilities required to securely keep the drone on the Spot independently whether it is standing or moving. Thus, in the designed unification solution, the landing platform does not impact the overall robot mobility and has no interference with the robot’s legs. The initial solution design was extensively evaluated in a series of tests at the laboratory, which demonstrated its viability.