Formal methods for control synthesis in partially observed environments : application to autonomous robotic manipulation

Modern robots are increasingly expected to function in uncertain and dynamically challenging environments, often in proximity with humans. In addition, wide scale adoption of robots requires on-the-fly adaptability of software for diverse application. These requirements strongly suggest the need to adopt formal representations of high level goals and safety specifications, especially as temporal logic formulas. This approach allows for the use of formal verification techniques for controller synthesis that can give guarantees for safety and performance. Robots operating in unstructured environments also face limited sensing capability. Correctly inferring a robot's progress toward high level goal can be challenging.

This thesis develops new algorithms for synthesizing discrete controllers in partially known environments under specifications represented as linear temporal logic (LTL) formulas. It is inspired by recent developments in finite abstraction techniques for hybrid systems and motion planning problems. The robot and its environment is assumed to have a finite abstraction as a Partially Observable Markov Decision Process (POMDP), which is a powerful model class capable of representing a wide variety of problems. However, synthesizing controllers that satisfy LTL goals over POMDPs is a challenging problem which has received only limited attention.

This thesis proposes tractable, approximate algorithms for the control synthesis problem using Finite State Controllers (FSCs). The use of FSCs to control finite POMDPs allows for the closed system to be analyzed as finite global Markov chain. The thesis explicitly shows how transient and steady state behavior of the global Markov chains can be related to two different criteria with respect to satisfaction of LTL formulas. First, the maximization of the probability of LTL satisfaction is related to an optimization problem over a parametrization of the FSC. Analytic computation of gradients are derived which allows the use of first order optimization techniques.

The second criterion encourages rapid and frequent visits to a restricted set of states over infinite executions. It is formulated as a constrained optimization problem with a discounted long term reward objective by the novel utilization of a fundamental equation for Markov chains - the Poisson equation. A new constrained policy iteration technique is proposed to solve the resulting dynamic program, which also provides a way to escape local maxima.

The algorithms proposed in the thesis are applied to the task planning and execution challenges faced during the DARPA Autonomous Robotic Manipulation - Software challenge.

Convex Relaxation for Low-Dimensional Representation: Phase Transitions and Limitations

There is a growing interest in taking advantage of possible patterns and structures in data so as to extract the desired information and overcome the curse of dimensionality. In a wide range of applications, including computer vision, machine learning, medical imaging, and social networks, the signal that gives rise to the observations can be modeled to be approximately sparse and exploiting this fact can be very beneficial. This has led to an immense interest in the problem of efficiently reconstructing a sparse signal from limited linear observations. More recently, low-rank approximation techniques have become prominent tools to approach problems arising in machine learning, system identification and quantum tomography.

In sparse and low-rank estimation problems, the challenge is the inherent intractability of the objective function, and one needs efficient methods to capture the low-dimensionality of these models. Convex optimization is often a promising tool to attack such problems. An intractable problem with a combinatorial objective can often be "relaxed" to obtain a tractable but almost as powerful convex optimization problem. This dissertation studies convex optimization techniques that can take advantage of low-dimensional representations of the underlying high-dimensional data. We provide provable guarantees that ensure that the proposed algorithms will succeed under reasonable conditions, and answer questions of the following flavor:

More specifically, i) Focusing on linear inverse problems, we generalize the classical error bounds known for the least-squares technique to the lasso formulation, which incorporates the signal model. ii) We show that intuitive convex approaches do not perform as well as expected when it comes to signals that have multiple low-dimensional structures simultaneously. iii) Finally, we propose convex relaxations for the graph clustering problem and give sharp performance guarantees for a family of graphs arising from the so-called stochastic block model. We pay particular attention to the following aspects. For i) and ii), we aim to provide a general geometric framework, in which the results on sparse and low-rank estimation can be obtained as special cases. For i) and iii), we investigate the precise performance characterization, which yields the right constants in our bounds and the true dependence between the problem parameters.

Análise global da estabilidade termodinâmica de misturas: um estudo com o método do conjunto gerador

O cálculo do equilíbrio de fases é um problema de grande importância em processos da engenharia, como, por exemplo, na separação por destilação, em processos de extração e simulação da recuperação terciária de petróleo, entre outros. Mas para resolvê-lo é aconselhável que se estude a priori a estabilidade termodinâmica do sistema, a qual consiste em determinar se uma dada mistura se apresenta em uma ou mais fases. Tal problema pode ser abordado como um problema de otimização, conhecido como a minimização da função distância do plano tangente à energia livre de Gibbs molar, onde modelos termodinâmicos, de natureza não convexa e não linear, são utilizados para descrevê-lo. Esse fato tem motivado um grande interesse em técnicas de otimização robustas e eficientes para a resolução de problemas relacionados com a termodinâmica do equilíbrio de fases. Como tem sido ressaltado na literatura, para proporcionar uma completa predição do equilíbrio de fases, faz-se necessário não apenas a determinação do minimizador global da função objetivo do teste de estabilidade, mas também a obtenção de todos os seus pontos estacionários. Assim, o desenvolvimento de metodologias para essa tarefa desafiadora tem se tornado uma nova área de pesquisa da otimização global aplicada à termodinâmica do equilíbrio, com interesses comuns na engenharia química e na engenharia do petróleo. O foco do presente trabalho é uma nova metodologia para resolver o problema do teste de estabilidade. Para isso, usa-se o chamado método do conjunto gerador para realizar buscas do tipo local em uma rede de pontos previamente gerada por buscas globais efetuadas com uma metaheurística populacional, no caso o método do enxame de partículas.Para se obter mais de um ponto estacionário, minimizam-se funções de mérito polarizadas, cujos pólos são os pontos previamente encontrados. A metodologia proposta foi testada na análise de quatorze misturas polares previamente consideradas na literatura. Os resultados mostraram que o método proposto é robusto e eficiente a ponto de encontrar, além do minimizador global, todos os pontos estacionários apontados previamente na literatura, sendo também capaz de detectar, em duas misturas ternárias estudadas, pontos estacionários não obtidos pelo chamado método de análise intervalar, uma técnica confiável e muito difundida na literatura. A análise do teste de estabilidade pela simples utilização do método do enxame de partículas associado à técnica de polarização mencionada acima, para a obtenção de mais de um ponto estacionário (sem a busca local feita pelo método do conjunto gerador em uma dada rede de pontos), constitui outra metodologia para a resolução do problema de interesse. Essa utilização é uma novidade secundária deste trabalho. Tal metodologia simplificada exibiu também uma grande robustez, sendo capaz de encontrar todos os pontos estacionários pesquisados. No entanto, quando comparada com a abordagem mais geral proposta aqui, observou-se que tal simplificação pode, em alguns casos onde a função de mérito apresenta uma geometria mais complexa, consumir um tempo de máquina relativamente grande, dessa forma é menos eficiente.

NEW METHODOLOGY FOR FEEDER RECONFIGURATION IN DISTRIBUTION NETWORKS WITH DISTRIBUTED ENERGY RESOURCES

This doctoral Thesis defines and develops a new methodology for feeder reconfiguration in distribution networks with Distributed Energy Resources (DER). The proposed methodology is based on metaheuristic Ant Colony Optimization (ACO) algorithms. The methodology is called Item Oriented Ant System (IOAS) and the doctoral Thesis also defines three variations of the original methodology, Item Oriented Ant Colony System (IOACS), Item Oriented Max-min Ant System (IOMMAS) y Item Oriented Max-min Ant Colony System (IOACS). All methodologies pursue a twofold objective, to minimize the power losses and maximize DER penetration in distribution networks. The aim of the variations is to find the algorithm that adapts better to the present optimization problem, solving it most efficiently. The main feature of the methodology lies in the fact that the heuristic information and the exploitation information (pheromone) are attached to the item not to the path. Besides, the doctoral Thesis proposes to use feeder reconfiguration in order to increase the distribution network capacity of accepting a major degree of DER. The proposed methodology and its three variations have been tested and verified in two distribution networks well documented in the existing bibliography. These networks have been modeled and used to test all proposed methodologies for different scenarios with various DER penetration degrees.

Uma dedução dos critérios de multicriticalidade e o cálculo de pontos tricríticos via otimização global

Uma dedução dos critérios de multicriticalidade para o cálculo de pontos críticos de qualquer ordem representa a formalização de ideias utilizadas para calcular pontos críticos e tricríticos e ainda amplia tais ideias. De posse desta dedução pode-se compreender os critérios de tricriticalidade e, com isso, através de uma abordagem via problema de otimização global pode-se fazer o cálculo de pontos tricríticos utilizando um método numérico adequado de otimização global. Para evitar um excesso de custo computacional com rotinas numéricas utilizou-se aproximações na forma de diferenças finitas dos termos que compõem a função objetivo. Para simular a relação P v - T optou-se pela equação de estado cúbica de Peng-Robinson e pela regra clássica de fluidos de van der Vaals, para modelagem do problema também se calculou os tensores de ordem 2, 3, 4 e 5 da função do teste de estabilidade. Os resultados obtidos foram comparados com dados experimentais e por resultados obtidos com outros autores que utilizaram métodos numéricos, equação de estado ou abordagem diferente das utilizadas neste trabalho.

Formulação teórica dos fundamentos da otimização global topográfica com análise de desempenho e aplicações à estabilidade de fases de misturas termodinâmicas

Métodos de otimização que utilizam condições de otimalidade de primeira e/ou segunda ordem são conhecidos por serem eficientes. Comumente, esses métodos iterativos são desenvolvidos e analisados à luz da análise matemática do espaço euclidiano n-dimensional, cuja natureza é de caráter local. Consequentemente, esses métodos levam a algoritmos iterativos que executam apenas as buscas locais. Assim, a aplicação de tais algoritmos para o cálculo de minimizadores globais de uma função não linear,especialmente não-convexas e multimodais, depende fortemente da localização dos pontos de partida. O método de Otimização Global Topográfico é um algoritmo de agrupamento, que utiliza uma abordagem baseada em conceitos elementares da teoria dos grafos, a fim de gerar bons pontos de partida para os métodos de busca local, a partir de pontos distribuídos de modo uniforme no interior da região viável. Este trabalho tem dois objetivos. O primeiro é realizar uma nova abordagem sobre método de Otimização Global Topográfica, onde, pela primeira vez, seus fundamentos são formalmente descritos e suas propriedades básicas são matematicamente comprovadas. Neste contexto, propõe-se uma fórmula semi-empírica para calcular o parâmetro chave deste algoritmo de agrupamento, e, usando um método robusto e eficiente de direções viáveis por pontos-interiores, estendemos o uso do método de Otimização Global Topográfica a problemas com restrições de desigualdade. O segundo objetivo é a aplicação deste método para a análise de estabilidade de fase em misturas termodinâmicas,o qual consiste em determinar se uma dada mistura se apresenta em uma ou mais fases. A solução deste problema de otimização global é necessária para o cálculo do equilíbrio de fases, que é um problema de grande importância em processos da engenharia, como, por exemplo, na separação por destilação, em processos de extração e simulação da recuperação terciária de petróleo, entre outros. Além disso, afim de ter uma avaliação inicial do potencial dessa técnica, primeiro vamos resolver 70 problemas testes, e então comparar o desempenho do método proposto aqui com o solver MIDACO, um poderoso software recentemente introduzido no campo da otimização global.

Simulated Annealing: Rigorous finite-time guarantees for optimization on continuous domains

Simulated annealing is a popular method for approaching the solution of a global optimization problem. Existing results on its performance apply to discrete combinatorial optimization where the optimization variables can assume only a finite set of possible values. We introduce a new general formulation of simulated annealing which allows one to guarantee finite-time performance in the optimization of functions of continuous variables. The results hold universally for any optimization problem on a bounded domain and establish a connection between simulated annealing and up-to-date theory of convergence of Markov chain Monte Carlo methods on continuous domains. This work is inspired by the concept of finite-time learning with known accuracy and confidence developed in statistical learning theory.

A method for VVER-1000 fuel rearrangement optimization taking into account both fuel cladding durability and burnup

A method for VVER-1000 fuel rearrangement optimization that takes into account both cladding durability and fuel burnup and which is suitable for any regime of normal reactor operation has been established. The main stages involved in solving the problem of fuel rearrangement optimization are discussed in detail. Using the proposed fuel rearrangement efficiency criterion, a simple example VVER-1000 fuel rearrangement optimization problem is solved under deterministic and uncertain conditions. It is shown that the deterministic and robust (in the face of uncertainty) solutions of the rearrangement optimization problem are similar in principle, but the robust solution is, as might be anticipated, more conservative. © 2013 Elsevier B.V.

Low-rank optimization for distance matrix completion

This paper addresses the problem of low-rank distance matrix completion. This problem amounts to recover the missing entries of a distance matrix when the dimension of the data embedding space is possibly unknown but small compared to the number of considered data points. The focus is on high-dimensional problems. We recast the considered problem into an optimization problem over the set of low-rank positive semidefinite matrices and propose two efficient algorithms for low-rank distance matrix completion. In addition, we propose a strategy to determine the dimension of the embedding space. The resulting algorithms scale to high-dimensional problems and monotonically converge to a global solution of the problem. Finally, numerical experiments illustrate the good performance of the proposed algorithms on benchmarks. © 2011 IEEE.

Minimum maneuver time calculation using convex optimization

The problem of calculating the minimum lap or maneuver time of a nonlinear vehicle, which is linearized at each time step, is formulated as a convex optimization problem. The formulation provides an alternative to previously used quasi-steady-state analysis or nonlinear optimization. Key steps are: the use of model predictive control; expressing the minimum time problem as one of maximizing distance traveled along the track centerline; and linearizing the track and vehicle trajectories by expressing them as small displacements from a fixed reference. A consequence of linearizing the vehicle dynamics is that nonoptimal steering control action can be generated, but attention to the constraints and the cost function minimizes the effect. Optimal control actions and vehicle responses for a 90 deg bend are presented and compared to the nonconvex nonlinear programming solution. Copyright © 2013 by ASME.

Computational engineering design for micro-scale combustors

At present, optimisation is an enabling technology in innovation. Multi-objective and multi-disciplinary design tools are essential in the engineering design process, and have been applied successfully in aerospace and turbomachinery applications extensively. These approaches give insight into the design space and identify the trade-offs between the competing performance measures satisfying a number of constraints at the same time. It is anticipated here that the same benefits can be obtained for the design of micro-scale combustors. In this paper, a multi-disciplinary automated design optimisation system was developed for this purpose, which comprises a commercial computational fluid dynamics package and a multi-objective variant of the Tabu Search optimisation algorithm. The main objectives that are considered in this study are to optimise the main micro-scale combustor design characteristics and to satisfy manufacturability considerations from the very beginning of the whole design operation. Hydrogen-air combustion as well as 14 geometrical and 2 operational parameters are used to describe and model the design problem. Two illustrative test cases will be presented, in which the most important device operational requirements are optimised, and the efficiency of the developed optimisation system is demonstrated. The identification, assessment and suitability of the optimum design configurations are discussed in detail. Copyright © 2012 by ASME.

Enhancing and developing the practical optimisation capabilities and intelligence of automatic design software

In the modern engineering design cycle the use of computational tools becomes a neces- sity. The complexity of the engineering systems under consideration for design increases dramatically as the demands for advanced and innovative design concepts and engineering products is expanding. At the same time the advancements in the available technology in terms of computational resources and power, as well as the intelligence of the design software, accommodate these demands and make them a viable approach towards the chal- lenge of real-world engineering problems. This class of design optimisation problems is by nature multi-disciplinary. In the present work we establish enhanced optimisation capabil- ities within the Nimrod/O tool for massively distributed execution of computational tasks through cluster and computational grid resources, and develop the potential to combine and benefit from all the possible available technological advancements, both software and hardware. We develop the interface between a Free Form Deformation geometry manage- ment in-house code with the 2D airfoil aerodynamic efficiency evaluation tool XFoil, and the well established multi-objective heuristic optimisation algorithm NSGA-II. A simple airfoil design problem has been defined to demonstrate the functionality of the design sys- tem, but also to accommodate a framework for future developments and testing with other state-of-the-art optimisation algorithms such as the Multi-Objective Genetic Algorithm (MOGA) and the Multi-Objective Tabu Search (MOTS) techniques. Ultimately, heav- ily computationally expensive industrial design cases can be realised within the presented framework that could not be investigated before. © 2012 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc.

Enhancing and developing the practical optimisation capabilities and intelligence of automatic design software

In the modern engineering design cycle the use of computational tools becomes a necessity. The complexity of the engineering systems under consideration for design increases dramatically as the demands for advanced and innovative design concepts and engineering products is expanding. At the same time the advancements in the available technology in terms of computational resources and power, as well as the intelligence of the design software, accommodate these demands and make them a viable approach towards the challenge of real-world engineering problems. This class of design optimisation problems is by nature multi-disciplinary. In the present work we establish enhanced optimisation capabilities within the Nimrod/O tool for massively distributed execution of computational tasks through cluster and computational grid resources, and develop the potential to combine and benefit from all the possible available technological advancements, both software and hardware. We develop the interface between a Free Form Deformation geometry management in-house code with the 2D airfoil aerodynamic efficiency evaluation tool XFoil, and the well established multi-objective heuristic optimisation algorithm NSGA-II. A simple airfoil design problem has been defined to demonstrate the functionality of the design system, but also to accommodate a framework for future developments and testing with other state-of-the-art optimisation algorithms such as the Multi-Objective Genetic Algorithm (MOGA) and the Multi-Objective Tabu Search (MOTS) techniques. Ultimately, heavily computationally expensive industrial design cases can be realised within the presented framework that could not be investigated before. ©2012 AIAA.

A novel modeling approach for design studies of brushless doubly fed induction generator based on magnetic equivalent circuit

Brushless doubly fed induction generator (BDFIG) has substantial benefits, which make it an attractive alternative as a wind turbine generator. However, it suffers from lower efficiency and larger dimensions in comparison to DFIG. Hence, optimizing the BDFIG structure is necessary for enhancing its situation commercially. In previous studies, a simple model has been used in BDFIG design procedure that is insufficiently accurate. Furthermore, magnetic saturation and iron loss are not considered because of difficulties in determination of flux density distributions. The aim of this paper is to establish an accurate yet computationally fast model suitable for BDFIG design studies. The proposed approach combines three equivalent circuits including electric, magnetic and thermal models. Utilizing electric equivalent circuit makes it possible to apply static form of magnetic equivalent circuit, because the elapsed time to reach steady-state results in the dynamic form is too long for using in population-based design studies. The operating characteristics, which are necessary for evaluating the objective function and constraints values of the optimization problem, can be calculated using the presented approach considering iron loss, saturation, and geometrical details. The simulation results of a D-180 prototype BDFIG are compared with measured data in order to validate the developed model. © 1986-2012 IEEE.

Computational engineering design for micro-scale combustion devices: A thermally improved configuration

A multi-objective design optimisation study has been carried out with the objectives to improve the overall efficiency of the device and to reduce the fuel consumption for the proposed micro-scale combustor design configuration. In a previous study we identified the topology of the combustion chamber that produced improved behaviour of the device in terms of the above design criteria. We now extend our design approach, and we propose a new configuration by the addition of a micro-cooling channel that will improve the thermal behaviour of the design as previously suggested in literature. Our initial numerical results revealed an improvement of 2.6% in the combustion efficiency when we applied the micro-cooling channel to an optimum design configuration we identified from our earlier multi-objective optimisation study, and under the same operating conditions. The computational modelling of the combustion process is implemented in the commercial computational fluid dynamics package ANSYS-CFX using Finite Rate Chemistry and a single step hydrogen-air reaction. With this model we try to balance good accuracy of the combustion solution and at the same time practicality within the context of an optimisation process. The whole design system comprises also the ANSYS-ICEM CFD package for the automatic geometry and mesh generation and the Multi-Objective Tabu Search algorithm for the design space exploration. We model the design problem with 5 geometrical parameters and 3 operational parameters subject to 5 design constraints that secure practicality and feasibility of the new optimum design configurations. The final results demonstrate the reliability and efficiency of the developed computational design system and most importantly we assess the practicality and manufacturability of the revealed optimum design configurations of micro-combustor devices. Copyright © 2013 by ASME.