Collision avoidance is an essential capability for autonomous and assisted-driving ground vehicles. In this work, we developed a novel model predictive control based intelligent collision avoidance (CA) algorithm for a multi-trailer industrial ground vehicle implemented on a General Purpose Graphical Processing Unit (GPGPU). The CA problem is formulated as a multi-objective optimal control problem and solved using a limited look-ahead control scheme in real-time. Through hardware-in-the-loop-simulations and experimental results obtained in this work, we have demonstrated that the proposed algorithm, using NVIDA’s CUDA framework and the NVIDIA Jetson TX2 development platform, is capable of dynamically assisting drivers and maintaining the vehicle a safe distance from the detected obstacles on-thely. We have demonstrated that a GPGPU, paired with an appropriate algorithm, can be the key enabler in relieving the computational burden that is commonly associated with model-based control problems and thus make them suitable for real-time applications.

A cyclotron is used for diagnosing and treating cancer. Pipes in the cyclotron have to be replaced as they get worn out when isotopes travel through them. This thesis aims to use machine learning models to predict when these parts have to be changed. Based on previous studies for predictive maintenance three dif- ferent machine learning models are used. The chosen models are random forest, gradient boosting and support vector machine. The results show that a gradient boosting regressor that predicts the number of remaining runs before the pipes have to be changed in the cyclotron is preferred. However, some data augmenta- tion had to be done to obtain these results, and future studies could explore the possibility of using a bigger data set or a multiple classifier approach.

Telerobotics are becoming increasingly more prevalent in the medical field due to the many advantages they have over standard methods. In particular, their use in surgical procedures provides benefits such as safer operations and greater health care access, among others. However, there are drawbacks that dissuade telerobotic usage and aspects that can still be improved upon. Examples of these include the inevitable impact of time delay and the existence of disturbances, such as patient-manipulator contact. These factors can result in system destabilization and ultimately task failure, discouraging the usage of telerobotics in these settings. This thesis investigates the effects of time delay and contact disturbances on telerobotic performance in a surgical setting. In this work, different methods of improving telerobotic performance such as using hybrid controllers and predictive technology are explored. The goal is to investigate options for mitigating the negative effects of these elements while improving overall telerobotic performance.

A research report submitted in partial fulfillment of the requirements for the degree of Master of Science in School of Computer Science and Applied Mathematics to the Faculty of Science University of Witwatersrand, 2019 / Learning-based methods have recently become popular in control engineering, achieving good performance on a number of challenging tasks. However, in complex environments where data efficiency and safety are critical, current methods remain unsatisfactory. As a step toward addressing these shortcomings, we propose a learning-based approach that combines Gaussian process regression with model predictive control. Using sparse spectrum Gaussian processes, we extend previous work by learning a model of the dynamics incrementally from a stream ofsensory data. Utilizinglearned dynamics and model uncertainty, we develop a controller that can learn and plan in real-time under non-linear constraints. We test our approach on pendulum and cartpole swing up problems and demonstrate the benefits of learning on a challenging autonomous racing task. Additionally, we show that learned dynamics models can be transferred to new tasks without any additional training. / TL (2020)

This thesis investigates one option for improving the control of a reheating furnace used in heating steel slabs before hot rolling; an essential part of the steel manufacturing process. The furnace consumes a significant amount of energy, leading to high cost and high carbon dioxide emissions. The proposed solution is the implementation of a model predictive control (MPC) system to improve control and reduce fuel usage. The MPC system will be based on the use of system identification techniques to find a prediction model of the furnace, specifically using ARMAX models. An additional simulation model will be used to simulate the system, and to compare the performance of MPC and PID. The prediction model is found to have a normalized root mean squared error of over 91% for the first five minutes, suggesting that it has potential to be used for MPC. The simulation model has significant inaccuracies, due to the presence of unmeasured disturbances. The simulation results, although limited due to the inaccuracies of the simulation model, suggest that MPC is a viable option for improved control of the furnace. The use of MPC can potentially improve the repeatability of the heating process, resulting in improved steel quality and reduced defects. This thesis suggests that further investigation into the use of MPC for controlling reheating furnaces in the steel industry is worth pursuing, and could potentially bring significant benefits to both producers and the environment.

The purpose of this research was to investigate the relation between measurement bias at the item level (differential item functioning, dif) and predictive bias at the test score level. Dif was defined as a difference in the probability of getting a test item correct for examinees with the same ability but from different subgroups. Predictive bias was defined as a difference in subgroup regression intercepts and/or slopes in predicting a criterion. Data were simulated by computer. Two hypothetical subgroups (a reference group and a focal group) were used. The predictor was a composite score on a dimensionally complex test with 60 items. Sample size (35, 70, and 105 per group), validity coefficient (.3 or .5), and the mean difference on the predictor (0, .33, .66, and 1 standard deviation, sd) and the criterion (0 and .35 sd) were manipulated. The percentage of items showing dif (0%, 15%, and 30%) and the effect size of dif (small = .3, medium = .6, and large = .9) were also manipulated. Each of the 432 conditions in the 3 x 2 x 4 x 2 x 3 x 3 design was replicated 500 times. For each replication, a predictive bias analysis was conducted, and the detection of predictive bias against each subgroup was the dependent variable. The percentage of dif and the effect size of dif were hypothesized to influence the detection of predictive bias; hypotheses were also advanced about the influence of sample size and mean subgroup differences on the predictor and criterion. Results indicated that dif was not related to the probability of detecting predictive bias against any subgroup. Results were inconsistent with the notion that measurement bias and predictive bias are mutually supportive, i.e., the presence (or absence) of one type of bias is evidence in support of the presence (or absence) of the other type of bias. Sample size and mean differences on the predictor/criterion had direct and indirect effects on the probability of detecting predictive bias against both reference and focal groups. Implications for future research are discussed.

Engineered products can impact the day-to-day life of their users and other stakeholders. These impacts are often referred to as the product's social impacts. Products have been known to impact the people who use them, design them, manufacture them, distribute them, and the communities where they exist. Currently, there are few methods that can help an engineer identify, quantify, predict, or improve a product's social impact. Some companies and organizations have tried to identify their impacts and, for example, set goals for achieving more sustainable business practices. However, engineers, in large part, do not have methods that can help improve the sustainability and social impacts of their products. Without new methods to help engineers make better product decisions, products will continue to have unanticipated negative impacts and will likely not reach their true social impact potential. Engineers working in the field of Engineering for Global Development (EGD) are especially in need of methods that can help improve the social impacts of their products. One of the purposes of creating products in EGD is to help solve problems that lead to improved quality of life for people and communities in developing countries. The research in this dissertation presents new methods developed to help engineers understand, predict, and improve the social impact of their products. Chapter 2 introduces the Product Impact Metric, a simple metric engineers can use to quantify their products impact on improving the quality of life of impoverished individuals in developing countries. Chapter 3 introduces a method that engineers can use to create product-specific social impact metrics and models. These models are used to predict the social impacts of an expanded US-Mexico border wall on immigrants, border patrol officers, and local communities. Chapter 4 shows a method that allows engineers to create social impact models for individuals within a population. Using data available through online databanks and census reports, the author predicts the social impact of a new semi-automated cassava peeler on farmers in the Brazilian Amazon. In Chapter 5, the author presents a method for engineers to optimize a product according to its social impact on multiple stakeholders. Inspired by existing literature on multi-stakeholder decision making, eight different optimization problem formulations are presented and demonstrated in an example with the cassava peeler. Chapter 6 presents the author's experience in co-designing a semi-automated cassava with the Itacoatiara Rural Farming Cooperative. The peeler was designed and built by the author and is used as the example in Chapters 4 and 5. Finally, Chapter 7 shows the conclusions the author has in completing this research. Comments are made as to the difficulties encountered in this research (specifically data quality and validation), and the author makes suggestions of possible future work.

京都大学 / 新制・課程博士 / 博士(情報学) / 甲第24266号 / 情博第810号 / 新制||情||136(附属図書館) / 京都大学大学院情報学研究科システム科学専攻 / (主査)教授 大塚 敏之, 教授 石井 信, 教授 森本 淳 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

The increasing variety of data mining tools offers a large palette of types and representation formats for predictive models. Managing the models then becomes a big challenge, as well as reusing the models and keeping the consistency of model and data repositories. Sustainable access and quality assessment of these models become limited to researchers. The approach for the Data and Model Governance (DMG) makes easier to process and support complex solutions. In this thesis, contributions are proposed towards ensembles of models with a focus on model representation, comparison and usage. Predictive Toxicology was chosen as an application field to demonstrate the proposed approach to represent predictive models linked to data for DMG. Further analysing methods such as predictive models comparison and predictive models combination for reusing the models from a collection of models were studied. Thus in this thesis, an original structure of the pool of models was proposed to represent predictive toxicology models called Predictive Toxicology Markup Language (PTML). PTML offers a representation scheme for predictive toxicology data and models generated by data mining tools. In this research, the proposed representation offers possibilities to compare models and select the relevant models based on different performance measures using proposed similarity measuring techniques. The relevant models were selected using a proposed cost function which is a composite of performance measures such as Accuracy (Acc), False Negative Rate (FNR) and False Positive Rate (FPR). The cost function will ensure that only quality models be selected as the candidate models for an ensemble. The proposed algorithm for optimisation and combination of Acc, FNR and FPR of ensemble models using double fault measure as the diversity measure improves Acc between 0.01 to 0.30 for all toxicology data sets compared to other ensemble methods such as Bagging, Stacking, Bayes and Boosting. The highest improvements for Acc were for data sets Bee (0.30), Oral Quail (0.13) and Daphnia (0.10). A small improvement (of about 0.01) in Acc was achieved for Dietary Quail and Trout. Important results by combining all the three performance measures are also related to reducing the distance between FNR and FPR for Bee, Daphnia, Oral Quail and Trout data sets for about 0.17 to 0.28. For Dietary Quail data set the improvement was about 0.01 though, but this data set is well known as a difficult learning exercise. For five UCI data sets tested, similar results were achieved with Acc improvement between 0.10 to 0.11, closing more the gaps between FNR and FPR. As a conclusion, the results show that by combining performance measures (Acc, FNR and FPR), as proposed within this thesis, the Acc increased and the distance between FNR and FPR decreased.

A globular cluster is a collection of stars that orbits the center of its galaxy as a single satellite. Understanding what influences the formations of these clusters provides understanding of galaxy structure and insight into their early development. We continue the work of Harris et al. (2013), who identified a set of predictors that accurately determined the number of clusters Ngc, through analysis of an incomplete dataset. We aimed to improve upon these results through imputation of the missing data. A small amount of precision was gained for the slope of Ngc~ R_e*sigma_ e, while the intercept suffered a small loss of precision. Estimates of intrinsic variance also increased with the addition of imputed data. We also found galaxy morphological type to be a significant predictor of Ngc in a model with R_e*sigma_ e. Although it increased precision of the slope and reduced the residual variance, its overall contribution was negligible. / Thesis / Master of Science (MSc)