During the last decades, improved sensor and hardware technologies as well as new methods and algorithms have made self-driving vehicles a realistic possibility in the near future. Thanks to this technology enhancement, many leading automotive and technology companies have turned their attention towards developing advanced driver assistance systems (ADAS) and self-driving vehicles. Autonomous vehicles are expected to have their first big impact in closed areas, such as mines, harbors and loading/offloading sites. In such areas, the legal requirements are less restrictive and the surrounding environment is more controlled and predictable compared to urban areas. Expected positive outcomes include increased productivity and safety, reduced emissions and the possibility to relieve the human from performing complex or dangerous tasks. Within these sites, different truck and trailer systems are used to transport materials. These systems are composed of several interconnected modules, and are thus large and highly unstable while reversing. This thesis addresses the problem of designing efficient motion planning and feedback control frameworks for such systems. First, a cascade controller for a reversing truck with a dolly-steered trailer is presented. The unstable modes of the system is stabilized around circular equilibrium configurations using a gain-scheduled linear quadratic (LQ) controller together with a higher-level pure pursuit controller to enable path following of piecewise linear reference paths. The cascade controller is then used within a rapidly-exploring random tree (RRT) framework and the complete motion planning and control framework is demonstrated on a small-scale test vehicle. Second, a path following controller for a reversing truck with a dolly-steered trailer is proposed for the case when the obtained motion plan is kinematically feasible. The control errors of the system are modeled in terms of their deviation from the nominal path and a stabilizing LQ controller with feedforward action is designed based on the linearization of the control error model. Stability of the closed-loop system is proven by combining global optimization, theory from linear differential inclusions and linear matrix inequality techniques. Third, a systematic framework is presented for analyzing stability of the closed-loop system consisting of a controlled vehicle and a feedback controller, executing a motion plan computed by a lattice planner. When this motion planner is considered, it is shown that the closed-loop system can be modeled as a nonlinear hybrid system. Based on this, a novel method is presented for analyzing the behavior of the tracking error, how to design the feedback controller and how to potentially impose constraints on the motion planner in order to guarantee that the tracking error is bounded and decays towards zero. Fourth, a complete motion planning and control solution for a truck with a dolly-steered trailer is presented. A lattice-based motion planner is proposed, where a novel parametrization of the vehicle’s state-space is proposed to improve online planning time. A time-symmetry result is established that enhance the numerical stability of the numerical optimal control solver used for generating the motion primitives. Moreover, a nonlinear observer for state estimation is developed which only utilizes information from sensors that are mounted on the truck, making the system independent of additional trailer sensors. The proposed framework is implemented on a full-scale truck with a dolly-steered trailer and results from a series of field experiments are presented.

The purpose of this thesis was to evaluate currently commercially available head mounted virtual reality displays for potential use in pilot training simulators. For this purpose acommercial simulator was modified to display the virtual environment in an Oculus RiftDK2 headset. A typical monitor based setup was used to provide a set of hardware requirements which the VR implementation had to meet or exceed to be considered potentially usable for pilot training simulators. User tests were then performed with a group of users representative of those normally using pilot training simulators, including both pilots and engineers working with simulator development. The main focus of the user tests was to evaluate some potential weaknesses found in the technical comparison (such as when a measured parameter was close to the lower limit defined by the monitor based setup) and to make a measurement of the usability of the VR implementation. The results from the technical comparison showed that the technical requirements were met and in most cases also exceeded. There were however some potential weaknesses revealed during the user tests, which included screen resolution and the field of view. There was one main critical deficiency found during the user tests. This was the lack of interaction with the aircraft as users were only able to interact with the flight stick and throttle lever. While this enabled the users to control many aspects of the aircraft (by using buttons and other controls fitted on the flight stick/throttle) in a training scenario a user also has to be able to interact with other switches and/or monitors in the cockpit. This was however a known limitation of the implementation and thus didn’t affect the tested parts of the simulator. The user tests also confirmed that the resolution was a potential problem, but that the overall usability was high. Thus the VR implementation had potential for use in a pilot training simulator, if the critical issues found during the user tests were solved.

The consequences of climate change are already noticeable in small island developing states. Road networks are crucial for a functioning society, and are particularly vulnerable to extreme weather, floods, landslides and other effects of climate change. Road systems in small island developing states are therefore in special need of climate adaptation efforts. Climate adaptation of road systems also has to be cost-efficient since these small island states have limited economical resources. Recent advances in deep reinforcement learning, a subfield of artificial intelligence, has proven that intelligent agents can achieve superhuman level at a number of tasks, setting hopes high for possible future applications of the algorithms. To investigate wether deep reinforcement learning is suitable for climate adaptation of road maintenance systems a simulator has been set up, together with three deep reinforcement learning agents, and two non-intelligent agents for performance comparisons. The results of the project indicate that deep reinforcement learning is suitable for use in intelligent road maintenance systems for climate adaptation in small island developing states.

In direct reference to the saying “time is money”, nowadays scenario simulations play a key role in the tasks people perform. Optimizing the time length of tasks and synchronizing them properly is essential to increased profits in any line of business. In this thesis the Bergkvist-Insjön sawmill yard process will be computer simulated. As the number of trucks arriving at the sawmill is unknown, the unexpected arrival of trucks would produce a high pressure on internal resources and handling operations. The aim of this paper is to optimize the usage of the resources in the Bergkvist-Insjön sawmill, by running three different scenarios built based on the real system simulation. Scenario number three, in which a log stacker only has the tasks to unload the trucks and supply the measurement station, has been found most efficient. In the simulation of this scenario, the number of logs processed by the sawmill is the highest one. Also, the time spent by the log stackers between their tasks is the shortest one from all scenarios. The results of this thesis revealed that the most efficient improvement of the sawmill yard would be gained by a different tasks’ priority for the operating log stackers.

In the field of data mining, transfer learning is the method of transferring knowledge from one domain into another. Using reviews from prisjakt.se, a Swedish price comparison site, and hotels.com this work investigate how the similarities between domains affect the results of transfer learning when using recurrent neural networks. We test several different domains with different characteristics, e.g. size and lexical similarity. In this work only relatively similar domains were used, the same target function was sought and all reviews were in Swedish. Regardless, the results are conclusive; transfer learning is often beneficial, but is highly dependent on the features of the domains and how they compare with each other’s.

Over the recent years, the software industry has faced a steady increase in the number of exposed and exploited software vulnerabilities. With more software and devices being connected to the internet every day, the need for proactive security measures has never been more important. One promising new technology for making software more secure is fuzz testing. This automated testing technique is based around generating a large number of test cases with the intention of revealing dangerous bugs and vulnerabilities. In this thesis work, a new direction within grey-box fuzz testing is evaluated against previous work. The presented approach uses sampled probability data in order to guide the fuzz testing towards program states that are expected to be easy to reach and beneficial for the discovery of software vulnerabilities. Evaluation of the design shows that the suggested approach provides no obvious advantage over existing solutions, but also indicates that the performance advantage could be dependent on the structure of the system under test. However, analysis of the design itself highlights several design decisions that could benefit from more extensive research. While the design proposed in this thesis work is insufficient for replacing current state of the art fuzz testing software, it provides a solid foundation for future research within the field. With the many insights gained from the design and implementation work, this thesis work aims to both inspire others and showcase the challenges of creating a probability-based approach to grey-box fuzz testing.

Automatization is a desirable feature in many business areas. Manually extracting information from a physical object such as a receipt is something that can be automated to save resources for a company or a private person. In this paper the process will be described of combining an already existing OCR engine with a developed python script to achieve data extraction of valuable information from a digital image of a receipt. Values such as VAT, VAT%, date, total-, gross-, and net-cost; will be considered as valuable information. This is a feature that has already been implemented in existing applications. However, the company that I have done this project for are interested in creating their own version. This project is an experiment to see if it is possible to implement such an application using restricted resources. To develop a program that can extract the information mentioned above. In this paper you will be guided though the process of the development of the program. As well as indulging in the mindset, findings and the steps taken to overcome the problems encountered along the way. The program achieved a success rate of 86.6% in extracting the most valuable information: total cost, VAT% and date from a set of 53 receipts originated from 34 separate establishments.

The main goal of this thesis will be the design and implementation of a 2048-point FFT on an FPGA through the use of VHDL code.The FFT will use a butterfly Radix-2 architecture with focus on the comparison of the parameters between the system with different Worlengths, Coefficient Wordlengths and Symbol Error rates as well as different modulation types, comparing 64QAM and 256QAM for the 5Gsystem.This implementation will replace an FFT function block in a Matlab based open source 5G NR simulator based on the 3GPP 15 standard and simulate spectrum, MSE payload,and SER performance.

Today’s airplanes and modern cars are equipped with displays to communicate important information to the pilot or driver. These displays needs to be tested for safety reasons; displays that fail can be a huge safety risk and lead to catastrophic events. Today displays are tested by checking the output signals or with the help of a person who validates the physical display manually. However this technique is very inefficient and can lead to important errors being unnoticed. MindRoad AB is searching for a solution where validation of the display is made from a camera pointed at it, text and numbers will then be recognized using a computer vision algorithm and validated in a time efficient and accurate way. This thesis compares the three different text detection algorithms, EAST, SWT and Tesseract to determine the most suitable for continued work. The chosen algorithm is then optimized and the possibility to develop a program which meets MindRoad ABs expectations is investigated. As a result several algorithms were combined to a fully working program to detect and recognize text in industrial displays.

<p>Wireless Sensor Network is an emerging technology that has applicationsin Wireless Actuators, remote controlling, distribution of softwareupdates and distribution of parameters to sensor nodes. This projectwork basically covers the concept of macro-diversity. This is a situationin which several transmitters are used for transferring the same signal (inmulti-hop sensor networks) to check the increase in connected nodes orin network coverage. Transmitter macro-diversity increases the receivedsignal strength and thus increases the signal-to-noise ratio which resultsin a lower outage probability. To accomplish this task three differentstrategies have been simulated using thirteen different cases. Broadcastingis used when forming SFN of size one (strategy one) and uni-castingis used for forming SFNs of size two (strategy two) and size three (strategythree).In this project reference material has been gathered frombooks, journals and web sources; and MATLAB has been used as thesimulation tool in which codes are written in the M programming language.The algorithm works firstly by discovering all the nodes that areconnected directly with the Base Station through multi-hoping, afterwhich the second algorithm is applied to check how many more nodescan be reached by forming SFNs. A gain of up to 79% was observedusing strategy one and strategy two and up to 83% in strategy three.The results shows that strategy one (Forming SFNs using BroadcastingTechnique) is the best as more nodes can be reached (for different cases)than for the other two strategies (forming SFNs using uni-casting technique).</p>