Typescript (photocopy). / Digitized by Kansas Correctional Industries

This project is the implementation and testing of a path tracker for a car-sized Unmanned Ground Vehicle. The vehicle, a Toyota Land Cruiser, was provided by SSC and hadpreviously been modified for remote operation. The developed path tracker uses a "follow the carrot" algorithm and has been writtenin C using the Robot Operating Software (ROS) framework and has been integrated intothe vehicles existing ROS powered software. During the implementation, the Gazebo rigid body simulator was used to simulate a simplified vehicle. Integration with the real sensors was performed using a small-scalecar, both indoors with the aid of a Vicon motion capture system and outdoors utilizingonly sensors available to the full-size car. The small-scale tests showed promise, howeverwhen full-scale field tests were performed the results showed some problems and reasonsfor these are discussed.

草圖繪製是創意產業最早期的一個工序。研究發現，草圖不但表達了畫家的想法，在繪製的過程，草圖也能為畫家帶來新靈感，所以草圖繪製是不可缺少的工序。然而，原始的草圖包含許多不必要的筆觸，所以進行後期製作前，必須用線重新勾畫。傳統上，這個過程是費時和繁瑣的，畫家必須人手把每一條需要的線都重新勾畫出來。當應用到動畫創作的時候，由於涉及的草圖數量龐大，情況會變得更壊，所以有必要把整個過程自動化。現有的研究都視草圖勾勒為一個線條分組和曲線擬合的過程，他們會把相近而順暢的筆觸組合在一起，形成單一線條。然而，他們都忽略了視覺感知上的一個重要法則──格式塔理論中的閉合原理。格式塔理論是一個知名的心理學理論，解釋人類如何透過整合理解各種視覺元素。根據格式塔理論中的閉合原理，我們往往把各種分隔的視覺元素整合為一個封閉的形狀。在這篇論文中，我提出閉合原理比格式塔理論中的其他原理更能幫助我們理解草圖，從而提出了一種基於區域的方法來勾勒草圖。實驗結果發現，我的方法在保有草圖上封閉形狀的能力上比現有的方法更優勝。 / Sketching is the earliest stage of production in art and design. Sketches are useful in conveying and developing ideas. However, raw sketches contain unnecessary strokes and must be converted to neat and tidy drawings before moving onto later stage of production. Traditionally, this conversion process is time-consuming and tedious since it is performed stroke by stroke manually. The situation is even worse when it comes to animation production which involves a huge number of sketches, so there is a strong motivation to automate the conversion process. Existing works formulate the conversion process as stroke grouping and curve fitting processes, in which close and continuous strokes are grouped together to form single strokes in the resulting image. Nevertheless, previous works overlooked an important law of visual perception: the law of closure in Gestalt principles. Gestalt principles concluded from early visual perception studies demonstrate how human perceive visual elements as different groups of lines and shapes. It states that we tend to group elements into closed shape even when a gap exists. In this thesis, we utilize the idea of law of closure and propose a region-based approach to refine sketches. Experiment result shows that this method outperforms the existing methods in terms of the capability of preserving salient regions in sketches. / Detailed summary in vernacular field only. / Wong, Ka Wing. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2012. / Includes bibliographical references (leaves 59-61). / Abstracts also in Chinese. / Chapter 1 --- Introduction --- p.6 / Chapter 2 --- Background --- p.8 / Chapter 2.1 --- Role of Sketch in Art and Design --- p.8 / Chapter 2.2 --- Characteristics of Raw Sketches --- p.10 / Chapter 2.3 --- Related Works --- p.11 / Chapter 2.3.1 --- Methods for Raster Image --- p.12 / Chapter 2.3.2 --- Methods for Vector Image --- p.14 / Chapter 2.3.3 --- Deficiency --- p.17 / Chapter 3 --- Gestalt Principles and Its Application --- p.18 / Chapter 3.1 --- Introduction to Gestalt Principles --- p.18 / Chapter 3.2 --- Existing Computational Model of Gestalt Principles --- p.20 / Chapter 3.3 --- Gestalt Principles for Outline Tracing --- p.22 / Chapter 3.3.1 --- Similarity --- p.23 / Chapter 3.3.2 --- Proximity --- p.23 / Chapter 3.3.3 --- Continuity --- p.24 / Chapter 3.3.4 --- Regularity --- p.25 / Chapter 3.3.5 --- Closure --- p.25 / Chapter 4 --- Proposed Method --- p.27 / Chapter 4.1 --- Multi-scale Region Retrieval --- p.28 / Chapter 4.1.1 --- Construction of Region Hierarchy --- p.29 / Chapter 4.1.2 --- Region Refinement --- p.31 / Chapter 4.2 --- Salient Region Retrieval --- p.32 / Chapter 4.2.1 --- Flattening of Region Hierarchy --- p.32 / Chapter 4.2.2 --- Region Merging --- p.33 / Chapter 4.2.3 --- Region Pruning --- p.37 / Chapter 4.2.4 --- Region Merging by User --- p.41 / Chapter 4.3 --- Outline Synthesis --- p.43 / Chapter 4.3.1 --- Outline Synthesis of Region-boundary Strokes --- p.44 / Chapter 4.3.2 --- Smoothing of Region-boundary Strokes --- p.45 / Chapter 4.3.3 --- Outline Synthesis of Feature Strokes --- p.48 / Chapter 4.4 --- Curve Fitting --- p.50 / Chapter 5 --- Result and Discussion --- p.52 / Chapter 6 --- Conclusion --- p.58 / Chapter 7 --- Reference --- p.59

Many security systems employing different methods have been proposed to protect buried oil pipelines transporting petroleum products from the well head via the refinery to: depots and other receiving stations. Currently there is a security gap in the monitoring of these buried pipelines in real time and in keeping them protected from third party interference. This thesis addresses the problem of monitoring these systems by developing an automated image analysis system with the aid of a low-cost multisensory Unmanned Aerial Vehicle (UAV) for monitoring of buried pipeline right-of-way (ROW). The method used in this research is based on the identification of threat objects of interest from the video frame sequences of the pipeline right-of-way acquired by the UAV. This is achieved by training the system to recognise objects of interest using trained correlation filters. To determine the geographical location of detected objects, the Video frame sequences captured by the UAV platform were ortho-rectified to form ortho-images which were then mosaicked to form a seamless Digital Surface Model (DSM) covering the test area using a photogrammetry model. The DSM formed from the mosaicking of ortho-images is then emerged with a digital globe for geo-referencing of detected objects. Experiments were carried out on a test field located in United Kingdom and Nigeria, where video and telemetry data were collected, then processed using the techniques created in this research. The results demonstrated that the developed correlation filter was able to detect objects of interest despite the distortions that come with the object image, due to the fact that the expected distortion was compensated for using the training images. When compared with the 6 control points in the digital globe the accuracy of the two-dimension DSM gave a misalignment error of between 2 and 3 metres.

Energy efficiency is becoming an important objective for modern manufacturing industry. The aim of this work is to improve energy efficiency of an automated system. Since a majority of production processes are limited by an external bottleneck, the hypothesis of this work is that reducing the processing rate of the restricted processes can lead to saving in energy and resources. A methodology based on optimisation at process, cell and line levels is developed and evaluated over different scenarios.The developed methodology is then applied to a simulated production cell to study its efficacy quantitatively. In this particular case, the proposed approach yields a decrease in energy consumption of 49% at maximum production capacity. This decrease can be greater if there is an external factor such as low demand or another stage in the production line.

The vision of self driving cars has existed for a long time and the field of autonomous vehicles has been of great interest to researchers and companies. Volvo construction equipment presented their Electrical Site project in September 2016, with predictions of reducing carbon emission up to 95% and total cost of ownership by 25%. In the project, multiple autonomous haulers are intended to work in a fleet, loading, unloading and charging in a cyclic behavior. This masterthesis focus on the lateral control system of the automated hauler platform HX. The platform is modeled in an comprehensive simulation environment and three different control algorithms have been implemented and tested; An adaptive Proportional, Integral and Derivative (PID) controller, Stanley and the Proportional Integral + Proportional controller. The PID controller is tuned using the Nyquist stability criterion and the other two algorithms are tuned using a Genetic Algorithm. Results indicate that, to reach the optimal performance of the tested algorithms, manual tuning from experimental testing is required.

No description available.

Detta projekt handlar om att effektivisera EMC-testning genom att undersöka hur en modväxlare konstrueras samt hur denna styrs och ifall det behövs återkoppling för denna ska uppnå funktionell nivå. Syftet med modväxlare är att man enklare ska kunna testa emissions- och immunitetsnivåer av nya tekniska uppfinningar och på så sätt värna om framtidens elektromagnetiska milj¨o. Under arbetets g˚ang har en modv¨axlare designats med CAD varpå modellen har skalats ner för att testa positionsnogrannhet, acceleration och tidsåtgång för körning. Avslutningsvis har man dragit slutsatser baserat på testerna och med hjälp av dessa framfört anvisningar till hur en fullskalig modväxlare bör konstrueras.

This thesis investigates solutions to automate a lab process and give the customer basis for investment. An analysis of the current lab process and interviews with the staff were performed to set requirements for an automated solution. The customer wants to perform an analysis on a large number of combinations between antibiotics and bacteria. The issues that an automated solution is not able to handle were identified for the customer. A market analysis of existing solutions was preformed and contact was established with different suppliers. The suggested products were evaluated by performance and if they would be able to adapt to the process. The evaluation shows that with the same number of working hours the customer can increase the productivity with 6.7 times using a semi-automated system and a fully-automated system will result in an increase of 18.3 times. A guidance of how the customer will implement the system is described in the report. The report shows that an automated solution will make their process more efficient.

Capacitor discharge welding (CDW) is a rapid solidification joining process under the influence of one-dimensional thermal gradients. Although CDW is useful for joining small parts and dissimilar metals, CD welded joints have a large variability in weld strength. CDW is not widely accepted because of a lack of automated process control. Studying the sources of variability in the CDW process can guide the automation of CDW. Therefore, the objectives of this study were to investigate sources of variability that affect weld strength and to generate a model to predict the weld strength variability in CDW. The source of variability was investigated by using screening experiments. Four different materials, stainless steel, Nitronic 50 Steel, copper, and low oxygen copper (C101), were selected to represent various levels of thermal conductivity and absorbed gas content. Thermal conductivity, percentage of gas content absorbed, diameter, and welding time were treated as the independent variables while the dependent variables were the standard deviation of CD weld strength as a percentage of base material strength and the mean of CD weld strength as a percentage of base material strength. A screening experiment and a statistical analysis of the data were used to develop a predictive model of the weld strength variability in CDW. Electron photomicrographs of weld fracture surfaces and dynamic current and resistance curves for each welding cycle were used to support conclusions from the statistical analysis. Conclusions of this study are that thermal conductivity and absorbed gas content do have a significant influence on weld strength variability in CDW. / Graduation date: 1999