Mobile robots, either autonomous or tele-operated have the potential of assisting humans in various situations such as during natural disasters, Urban Search and Rescue (USAR) efforts, and in Explosive Ordinance Disposal (EOD). These robots need steady wireless connectivity with their base station for control and communication. On one hand, the wireless link has to be managed to maintain a stable high quality network connection. On other hand, wireless connection should be continuously monitored to foresee network failure or inadequate link quality situations caused by entering access with low signal strength. This thesis focus on the later where we aim to address the prediction of wireless network connectivity for mobile robots. To indicate wireless connection quality, we use the Radio Signal Strength (RSS) parameter which is readily available by most wireless devices, and it has been frequently used in the literature to indicate wireless connection quality as the RSS have direct relation to the network throughput. Thus the focus of this thesis is to predict the RSS in future robot positions with reference to the current position of the robot. The solution is not straight forward because of the challenging nature of the radio signal propagation which involves complex phenomena such as path loss, shadowing and multipath fading. The RSS prediction method designed in this thesis has two stages. In the first stage, we estimate the location of radio signal source using an RSS gradient-based approach that can work in both single and multiple receivers arrangements. This information will be applied in the next prediction stage. For RSS prediction, we make use of Gaussian Process Regression (GPR) due to non-parametric nature, robustness to noise in the RSS data and changes in the environment. We validate our design with extensive experiments conducted using different types of mobile robots and wireless devices in indoor and outdoor environments, and under line-of-sight (LOS) and non-line-of-sight (NLOS) conditions. We are able to achieve results with source localization error of up to 2 meters for indoor and 5 meters for outdoor environment. In terms of RSS prediction, we obtain the mean absolute prediction error of less than 5 dBm on average, for prediction within 5 meters in indoor environment and 20 meters in outdoor environment. The work is not only promising in terms of prediction time and accuracy but also outperform the state-of-the-art (SOTA) methods including the GPR algorithm, the Kriging interpolation method and the linear regression approaches.

No description available.

This thesis mainly consists of five independent papers concerning the reactive control design of autonomous mobile robots in the context of target tracking and cooperative formation keeping with obstacle avoidance in the static/dynamic environment. Technical contents of this thesis are divided into three parts. The first part consists of the first two papers, which consider the target-tracking and obstacle avoidance in the static environment. Especially, in the static environment, a fundamental issue of reactive control design is the local minima problem(LMP) inherent in the potential field methods(PFMs). Through introducing a state-dependent planned goal, the first paper proposes a switching control strategy to tackle this problem. The control law for the planned goal is presented. When trapped into local minima, the robot can escape from local minima by following the planned goal. The proposed control law also takes into account the presence of possible saturation constraints. In addition, a time-varying continuous control law is proposed in the second paper to tackle this problem. Challenges of finding continuous control solutions of LMP are discussed and explicit design strategies are then proposed. The second part of this thesis deals with target-tracking and obstacle avoidance in the dynamic environment. In the third paper, a reactive control design is presented for omnidirectional mobile robots with limited sensor range to track targets while avoiding static and moving obstacles in a dynamically evolving environment. Towards this end, a multiiii objective control problem is formulated and control is synthesized by generating a potential field force for each objective and combining them through analysis and design. Different from standard potential field methods, the composite potential field described in this paper is time-varying and planned to account for moving obstacles and vehicle motion. In order to accommodate a larger class of mobile robots, the fourth paper proposes a reactive control design for unicycle-type mobile robots. With the relative motion among the mobile robot, targets, and obstacles being formulated in polar coordinates, kinematic control laws achieving target-tracking and obstacle avoidance are synthesized using Lyapunov based technique, and more importantly, the proposed control laws also take into account possible kinematic control saturation constraints. The third part of this thesis investigates the cooperative formation control with collision avoidance. In the fifth paper, firstly, the target tracking and collision avoidance problem for a single agent is studied. Instead of directly extending the single agent controls to the multiagents case, the single agent controls are incorporated with the cooperative control design presented in [1]. The proposed decentralized control is reactive, considers the formation feedback and changes in the communication networks. The proposed control is based on a potential field method, its inherent oscillation problem is also studied to improve group transient performance.

[pt] Embora os sistemas multimídia tradicionais se concentrem na codificação e no armazenamento eficientes de tipos de mídia e suas relações temporais, a demanda atual por experiências mais ricas e personalizadas exige uma compreensão mais profunda do conteúdo semântico dessas mídias. Neste estudo, propomos a integração do processamento de nível semântico aos sistemas multimídia, enriquecendo o conteúdo com informações sobre entidades do mundo real, como objetos, ações, agentes e interpretação de linguagem. A principal contribuição desta dissertação é a apresentação de uma arquitetura para enriquecimento de dados multimídia em tempo real que usa técnicas de aprendizado de máquina para extrair representações semânticas incorporando as ao fluxos de dados multimídia como um serviço nativo e básico. Para demonstrar concretamente a proposta, implementamos dois casos de uso que servem como provas de conceito, mostrando a viabilidade da arquitetura e sua eficácia em cenários práticos. / [en] While traditional multimedia systems focused on efficient coding and storage of media types and their temporal relationships, the current demand for rich and customized experiences calls for a deeper understanding of semantic content. In this study, we propose the integration of semantic-level processing into multimedia systems, enriching content with information about real-world entities, such as objects, actions, agents, and language interpretation. The main contribution of this dissertation is the proposal of an architecture for real-time multimedia data enhancement that is able to use machine learning techniques to extract semantic representations and incorporating it into multimedia data streams as a native and basic service. To provide a concrete demonstration of the proposal, we implement two use cases that serve as proofs-of-concept, showing the feasibility of the architecture and showcasing its effectiveness in practical scenarios.

Industry 4.0 is a new phase of industrial growth that has been ushered in by the quick development of digital technologies like the Internet of Things (IoT), artificial intelligence (AI), and robots. Collaborative robotic products have appeared in this changing environment, enabling robots to collaborate with people in open workspaces. The paradigm changes away from autonomous robotics and toward collaborative human-robot interaction (HRI) has made it necessary to look at novel ways to improve output, effectiveness, and security. Many benefits, including more autonomy and flexibility, have been made possible by the introduction of Autonomous Mobile Robots (AMRs) and later Automated Guided Vehicles (AGVs) for material handling. However, this incorporation of robots into communal workspaces also brings up safety issues that must be taken into account. This thesis aims to address potential threats arising from the increasing automation in shopfloors and shared workplaces between AMRs and human operators by exploring the capabilities of Mixed Reality (MR) technologies. By harnessing MR's capabilities, the aim is to mitigate safety concerns and optimize the effectiveness of collaborative environments. To achieve this the research is structured around the following sub-objectives: the development of a communication network enabling interaction among all devices in the shared workspace and the creation of a MR user interface promoting accessibility for human operators. A comprehensive literature review was conducted to analyse existing proposals aimed at improving HRI through various techniques and approaches. The objective was to leverage MR technologies to enhance collaboration and address safety concerns, thereby ensuring the smooth integration of AMRs into shared workspaces. While the literature review revealed limited research utilizing MR for data visualization in this specific domain, the goal of this thesis was to go beyond existing solutions by developing a comprehensive approach that prioritizes safety and facilitates operator adaptation. The research findings highlight the superiority of MR in displaying critical information regarding robot intentions and identifying safe zones with reduced AMR activity. The utilization of HoloLens 2 devices, known for their ergonomic design, ensures operator comfort during extended use while enhancing the accuracy of tracking positions and intentions in highly automated environments. The presented information is designed to be concise, customizable, and easily comprehensible, preventing information overload for operators.  The implementation of MR technologies within shared workspaces necessitates ethical considerations, including transparent data collection and user consent. Building trust is essential to establish MR as a reliable tool that enhances operator working conditions and safety. Importantly, the integration of MR technologies does not pose a threat to job displacement but rather facilitates the smooth adaptation of new operators to collaborative environments. The implemented features augment existing safety protocols without compromising efficacy, resulting in an overall improvement in safety within the collaborative workspace. In conclusion, this research showcases the effectiveness of MR technologies in bolstering HRI, addressing safety concerns, and enhancing operator working conditions within collaborative shopfloor environments. Despite encountering limitations in terms of time, complexity, and available information, the developed solution showcases the potential for further improvements. The chosen methodology and philosophical paradigm have successfully attained the research objectives, and crucial ethical considerations have been addressed. Ultimately, this thesis proposes and provides a comprehensive explanation for potential future implementations, aiming to expand the actual capabilities of the solution.

No description available.

No description available.

Autonomous vehicles are the subject of intense research because they are a safe and convenient alternative to present-day vehicles. Human drivers base their navigational decisions primarily on visual information and researchers have been attempting to use computers to do the same. The current challenge in using computer vision lies not in the collection or transmission of visual data, but in the perception of visual data to extract from it useful information. The focus of this thesis is on the use of computer vision to navigate an autonomous vehicle that will participate in the Intelligent Ground Vehicle Competition (IGVC.) This document starts with a description of the IGVC and the software design of an autonomous vehicle. This thesis then focuses on the weakest link in the system - the computer vision module. Vehicles at the IGVC are expected to autonomously navigate an obstacle course. Competing vehicles need to recognize and stay between lines painted on grass or pavement. The research presented in this document describes two methods used for boundary line extraction: color-based object extraction, and shape analysis for line recognition. This is the first time a combination of these methods is being applied to the problem of line recognition in the context of the IGVC. The most significant contribution of this work is a method for extracting lines in a binary image even when the line is attached to a shape that is not a line. Novel methods have been used to simplify camera calibration, and for perspective correction of the image. The results give promise of vastly improved autonomous vehicle performance. / Master of Science

Students at Virginia Tech have been developing autonomous vehicles for the past five years. The purpose of these vehicles has been primarily for entry in the annual international Intelligent Ground Vehicle Competition (IGVC), however further applications for autonomous vehicles range from UneXploded Ordinance (UXO) detection and removal to planetary exploration. Recently, Virginia Tech developed a successful autonomous vehicle named Navigator. Navigator was developed primarily for entry in the IGVC, but also intended for use as a research platform. For navigation, Navigator uses a local obstacle avoidance method known as the Vector Field Histogram (VFH). However, in order to form a complete navigation scheme, the local obstacle avoidance algorithm must be coupled with a global map. This work presents a simple algorithm for developing a quasi-free space global map. The algorithm is based on the premise that the robot will be given multiple attempts at a particular goal. During early attempts, Navigator explores using solely local obstacle avoidance. While exploring, Navigator records where it has been and uses this information on subsequent attempts. Further, this thesis outlines the look-ahead method by which the global map is implemented. Finally, both simulated and experimental results are presented. The aforementioned global map building algorithm uses a common method of localization known as odometry. Odometry, also referred to as dead reckoning, is subject to inaccuracy caused by systematic and non-systematic errors. In many cases, the most dominant source of inaccuracy is systematic errors. Systematic errors are inherent to the vehicle; therefore, the dead reckoning inaccuracy grows unbounded. Fortunately, it is possible to largely eliminate systematic errors by calibrating the parameters such that the differences between the nominal dimensions and the actual dimensions are minimized. This work presents a method for calibration of mobile robot parameters using optimization. A cost function is developed based on the well-known UMBmark (University of Michigan Benchmark) test pattern. This method is presented as a simple time efficient calibration tool for use during startup procedures of a differentially driven mobile robot. Results show that this tool consistently gives greater than 50% improvement in overall dead reckoning accuracy on an outdoor mobile robot. / Master of Science

The implementation of deep reinforcement learning in mobile robotics offers a great solution for the development of autonomous mobile robots to efficiently complete tasks and transport objects. Reinforcement learning continues to show impressive potential in robotics applications through self-learning and biological plausibility. Despite its advancements, challenges remain in applying these machine learning techniques in dynamic environments. This thesis explores the performance of Deep Q-Networks (DQN), using images as an input, for mobile robot navigation in dynamic maze puzzles and aims to contribute to advancements in deep reinforcement learning applications for simulated and real-life robotic systems. This project is a step towards implementation in a hardware-based system. The proposed approach uses a DQN algorithm with experience replay and an epsilon-greedy annealing schedule. Experiments are conducted to train DQN agents in static and dynamic maze environments, and various reward functions and training strategies are explored to optimize learning outcomes. In this context, the dynamic aspect involves training the agent on fixed mazes and then testing its performance on modified mazes, where obstacles like walls alter previously optimal paths to the goal. In game play, the agent achieved a 100\% win rate in both 4x4 and 10x10 static mazes, successfully making it to the goal regardless of slip conditions. The number of rewards obtained during the game-play episodes indicates that the agent took the optimal path in all 100 episodes of the 4x4 maze without the slip condition, whereas it took the shortest, most optimal path in 99 out of 100 episodes in the 4x4 maze with the slip condition. Compared to the 4x4 maze, the agent more frequently chose sub-optimal paths in the larger 10x10 maze, as indicated by the amount of times the agent maximized rewards obtained. In the 10x10 static maze game-play, the agent took the optimal path in 96 out of 100 episodes for the no slip condition, while it took the shortest path in 93 out of 100 episodes for the slip condition. In the dynamic maze experiment, the agent successfully solved 7 out of 8 mazes with a 100\% win rate in both original and modified maze environments. The results indicate that adequate exploration, well-designed reward functions, and diverse training data significantly impacted both training performance and game play outcomes. The findings suggest that DQN approaches are plausible solutions to stochastic outcomes, but expanding upon the proposed method and more research is needed to improve this methodology. This study highlights the need for further efforts in improving deep reinforcement learning applications in dynamic environments.