Dimitrios Hristu-Varsakelis's research while affiliated with University of Macedonia and other places

Publications (81)

Motivated by the persistent phenomenon of tax evasion and the challenge of tax collection during economic crises, we explore the behavior of a risk-neutral self-interested firm that may engage in tax evasion to maximize its profits. The firm evolves in a tax system which includes many of “standard” features such as audits, penalties, and occasional tax amnesties, and may be uncertain as to its tax status (not knowing, for example, whether a tax amnesty may be imminent). We show that the firm’s dynamics can be expressed via a partially observable Markov decision process and use that model to compute the firm’s optimal behavior and expected long-term discounted rewards in a variety of scenarios of practical interest. Going beyond previous work, we are able to investigate the effect of “leaks” or “pre-announcements” of any tax amnesties on the firm’s behavior (and thus on tax revenues). We also compute the effect on firm behavior of any extensions of the statute of limitations within which the firm’s tax filings can be audited, and show that such extensions can be a significant deterrent against tax evasion.
We present a deep long short-term memory (LSTM)-based neural network for predicting asset prices, together with a successful trading strategy for generating profits based on the model’s predictions. Our work is motivated by the fact that the effectiveness of any prediction model is inherently coupled to the trading strategy it is used with, and vise versa. This highlights the difficulty in developing models and strategies which are jointly optimal, but also points to avenues of investigation which are broader than prevailing approaches. Our LSTM model is structurally simple and generates predictions based on price observations over a modest number of past trading days. The model’s architecture is tuned to promote profitability, as opposed to accuracy, under a strategy that does not trade simply based on whether the price is predicted to rise or fall, but rather takes advantage of the distribution of predicted returns, and the fact that a prediction’s position within that distribution carries useful information about the expected profitability of a trade. The proposed model and trading strategy were tested on the S&P 500, Dow Jones Industrial Average (DJIA), NASDAQ and Russell 2000 stock indices, and achieved cumulative returns of 340%, 185%, 371% and 360%, respectively, over 2010-2018, far outperforming the benchmark buy-and-hold strategy as well as other recent efforts.
Designing tax policies that are effective in curbing tax evasion and maximize state revenues requires a rigorous understanding of taxpayer behavior. This work explores the problem of determining the strategy a self-interested, risk-averse tax entity is expected to follow, as it "navigates" - in the context of a Markov Decision Process - a government-controlled tax environment that includes random audits, penalties and occasional tax amnesties. Although simplified versions of this problem have been previously explored, the mere assumption of risk-aversion (as opposed to risk-neutrality) raises the complexity of finding the optimal policy well beyond the reach of analytical techniques. Here, we obtain approximate solutions via a combination of Q-learning and recent advances in Deep Reinforcement Learning. By doing so, we i) determine the tax evasion behavior expected of the taxpayer entity, ii) calculate the degree of risk aversion of the "average" entity given empirical estimates of tax evasion, and iii) evaluate sample tax policies, in terms of expected revenues. Our model can be useful as a testbed for "in-vitro" testing of tax policies, while our results lead to various policy recommendations.
This work explores the use of a pen-and-tablet device to study differences in hand movement and muscle coordination between healthy subjects and Parkinson’s disease patients. We let volunteers draw simple horizontal lines and recorded the trajectory of the pen’s tip on the pad’s surface. The signals thus obtained were then processed to compute various features which correspond to the variability of the pen tip’s velocity, the deviation from the horizontal plane, and the trajectory’s entropy. Our goal was to establish simple and objective metrics which can be used to differentiate between normal and pathological movement. In a small-scale clinical trial, 44 age-matched subjects were divided in two groups, namely 20 healthy subjects (H), and 24 Parkinson’s disease (PD) patients. We applied a comprehensive machine learning approach to build a model that could classify unknown subjects based on their line-drawing performance. We were able to achieve an average prediction accuracy of 91% (88% sensitivity [ΤP], 95% specificity [ΤN]). Our results show that the proposed method is a good candidate for differentiating between healthy and Parkinson’s disease individuals, and shows promise in the context of telemedicine applications and tracking of the disease’s symptoms via inexpensive, widely available hardware.
The aim of this study is to propose a practical smartphone-based tool to accurately assess upper limb tremor in Parkinson's Disease (PD) patients. The tool uses signals from the phone's accelerometer and gyroscope (as the phone is held or mounted on a subject's hand) to compute a set of metrics which can be used to quantify a patient's tremor symptoms. In a smallscale clinical study with 25 PD patients and 20 age-matched healthy volunteers, we combined our metrics with machine learning techniques to correctly classify 82% of the patients and 90% of the healthy volunteers, which is high compared to similar studies. The proposed method could be effective in assisting physicians in the clinic, or to remotely evaluate the patient's condition and communicate the results to the physician. Our tool is low-cost, platform independent, non-invasive, and requires no expertise to use. It is also well-matched to the standard clinical examination for PD and can keep the patient "connected" to his physician on a daily basis. Finally, it can facilitate the creation of anonymous profiles for PD patients, aiding in further research on the effectiveness of medication or other overlooked aspects of patients' lives.
Abstract— With an ever-growing number of technologically advanced methods for the diagnosis and quantification of movement disorders, comes the need to assess their accuracy and see how they match up with widely used standard clinical assessment tools. This work compares quantitative measurements of hand tremor in twenty-three Parkinson’s disease patients, with their clinical scores in the hand tremor components of the Unified Parkinson’s Disease Rating Scale (UPDRS), which is considered the “gold standard” in the clinical assessment of the disease. Our measurements were obtained using a smartphone-based platform, which processes the phone’s accelerometer and gyroscope signals to detect and measure hand tremor. The signal metrics used were mainly based on the magnitude of the acceleration and the rotation rate vectors of the device. Our results suggest relatively strong correlation (r>0.7 and p<0.01) between the patients’ UPDRS hand tremor scores and the signal metrics applied to the measured signals.
We develop a Markov-based optimization model that captures the process via which a risk-averse firm in Greece decides whether to engage in tax evasion. The firm seeks to maximize the expected utility of its wealth, the latter viewed as a function of the portion of profits which the firm attempts to conceal from the government. Our model takes into account the basic features of the Greek tax system, including random audits and tax penalties applied when the audit reveals any wrongdoing. The proposed model is used to i) show that the parameters currently in place are conducive to tax evasion, and ii) "chart" the problem's parameter space in order to identify "virtuous" combinations (from the point of view of the government), and obtain a relationship between audit probability, tax penalty and likelihood of the firm engaging in tax evasion
One of the important but often overlooked practical challenges in motion control for robotics and other autonomous machines has to do with the implementation of theoretical tools into software that will allow the system to interact effectively with the physical world. More often than not motion control programs are machine-specific and not reusable, even when the underlying algorithm does not require any changes. The work on Motion Description Languages (MDL) has been an effort to formalize a general purpose robot programming language that allows one to incorporate both switching logic and differential equations. Extended MDL (MDLe) is a device-independent programming language for hybrid motion control, accommodating hybrid controllers, multi-robot interactions and robot-to-robot communications. The purpose of this paper is to describe the "MDLe engine", a software tool that implements the MDLe language. We have designed a basic compiler/software foundation for writing MDLe code. We provide a brief description of the MDLe syntax, implementation architecture, and functionality. Sample programs are presented together with the results of their execution on a set of physical and simulated mobile robots.
We present a Markov-based model of the process via which a ‘representative’ Greek risk-averse firm decides the degree to which it should engage in tax evasion. The model is constructed around a simplified version of the Greek tax system which includes random audits and penalties for under-reporting profits. For its part, the firm is allowed to manipulate its stated profits, potentially exposing itself to future penalty payments, in an attempt to maximize the expected utility of its after-tax wealth. Using our model, we determine the optimal behaviour expected of the firm as a function of the parameters of the tax system, and identify subsets of the audit probability – tax penalty space which ‘remove’ the inventive for tax evasion. This allows us to – among other things – evaluate the effectiveness of the parameter values currently in use and determine the implied level of risk-aversion for the average Greek firm.
In the midst of the financial crisis currently unfolding in Greece, tax revenue collection is considered a top priority. This work describes a dynamic, Markov-based decision support model, aimed at predicting the behavior of a risk-neutral enterprise in Greece, and at evaluating tax policies before they are implemented. We use our model to i) analyze the effectiveness of an alternative taxation option periodically offered by the Greek government, ii) show that in the current environment, a rational enterprise has no incentive to disclose its profits, and iii) identify “virtuous” combinations of parameters which lead to full disclosure of profits.
We explore an input–output based framework for optimizing production in the Greek economy, under constraints relating to energy use, final demand, greenhouse gas emissions and solid waste. Using empirical data, we consider the effects on the maximum attainable gross value of production when imposing various pollution abatement targets. Our results quantify those effects as well as the magnitude of economic sacrifices required to achieve environmental goals, in a series of policy scenarios of practical importance. Because air pollution and solid waste are not produced independently of one another, we identify the settings in which it is meaningful to institute a separate policy for mitigating each pollutant, versus those in which only one pollutant needs to be actively addressed. The scenarios considered here represent a range of options that could be available to policy makers, depending on the country's international commitments and the effects on economic and environmental variables.
Recent advances in mobile phone technology have placed an impressive array of sensing and communication equipment at the hands of an ever-growing number of people. One of the areas which can potentially be transformed by the availability of what is essentially a cheap, ubiquitous networked sensor, is that of remote diagnosis of movement disorders, such as Parkinson's disease. This work describes a smartphone-based method for detecting and quantifying the hand tremor associated with movement disorders using signals from the accelerometer and gyroscope embedded in the patient's phone. Our approach is web-based and user-friendly, requiring minimal user interaction. In clinical experiments with twenty subjects, we found that by combining both accelerometer and gyroscope signals, we were able to correctly identify those with hand tremor, using very simple signal metrics.
We consider a collection of countries which attempt to maximize their corporate tax revenue, the latter being viewed as a function of Foreign Direct Investment (FDI) inflow and the Effective Average Tax Rate (EATR) which each country sets for itself. Under a model that assumes a direct influence of tax differentials on the flow of FDI, each country's decisions are naturally [`]coupled' to those of others, leading to a non-cooperative game in which countries-players compete for FDI inflows by sequentially altering their tax rates. Their decisions are made via a differential equation-based model used to predict the effect of tax rate changes on a player's share of FDI inflows. Our model, calibrated using empirical data from 12 OECD countries for the period 1982-2005, combines FDI inflow and tax-rate differentials to arrive at a "steady-state" FDI inflow share for each player, given its competitors' corporate tax rates. We explore the game's equilibrium, including the question of whether equilibrium necessarily implies a [`]race to bottom', with low corporate tax rates for all players.
We discuss a recently proposed one-pass authenticated key agreement protocol, by Mohammad, Chen, Hsu and Lo, which was "derived" from their correponding two-pass version and claimed to be secure. We show that this is not the case by demonstrating a number of vulnerabilities.
Under its Kyoto and EU obligations, Greece has committed to a greenhouse gas (GHG) emissions increase of at most 25% compared to 1990 levels, to be achieved during the period 2008–2012. Although this restriction was initially regarded as being realistic, information derived from GHG emissions inventories shows that an increase of approximately 28% has already taken place between 1990 and 2005, highlighting the need for immediate action. This paper explores the reallocation of production in Greece, on a sector-by-sector basis, in order to meet overall demand constraints and GHG emissions targets. We pose a constrained optimization problem, taking into account the Greek environmental input–output matrix for 2005, the amount of utilized energy and pollution reduction options. We examine two scenarios, limiting fluctuations in sectoral production to at most 10% and 15%, respectively, compared to baseline (2005) values. Our results indicate that (i) GHG emissions can be reduced significantly with relatively limited effects on GVP growth rates, and that (ii) greater cutbacks in GHG emissions can be achieved as more flexible production scenarios are allowed.
An increasing number of recent articles applying powerful tests for non-linear causality have fuelled interest in discovering complex dynamics in macroeconomic and financial data. This paper is an attempt to add to the set of available tools by proposing a test for determining the source of causal relationships when a certain class of complex dynamics, that includes chaotic non-linearities, is present.
Key establishment protocols are among the most important security mechanisms via which two or more parties can encrypt their communications over an insecure network. This paper is concerned with the vulnerability of one-pass two-party key establishment protocols to key-compromise impersonation (K-CI) attacks. The latter may occur once an adversary has obtained the long-term private key of an honest party, and represent a serious — but often underestimated — threat, because a successful impersonation attack may result in far greater harm than the reading of past and future conversations. Our aim is to describe two main classes of K-CI attacks that can be mounted against all of the best-known one-pass protocols, including MQV and HMQV. We show that one of the attacks described can be somewhat avoided (though not completely eliminated) through the combined use of digital signatures and time-stamps; however, there still remains a class of K-CI threats for which there is no obvious solution.
As online transactions become increasingly practical, a broad range of industrial and e-government applications have emerged which depend on time-based protection of confidential data. Despite theoretical progress in timed-release cryptography (TRC), there is still no implementation infrastructure that takes advantage of the latest TRC algorithms. The purpose of this paper is to propose such an infrastructure for pairing-based timed-release cryptography (PB-TRC) systems. Our model uses key generation centers (KGCs) which publish decryption keys periodically, and satisfies the security requirements of modern third party-based TRC schemes. Our approach combines the best features of existing models into a generic and complete infrastructure which is to support TRC. It is also "lighter" in terms of complexity and communication, and is as effective (in terms of security and related properties) as the TRC protocol it is used with.
We explore the LQG control of a networked control system (NCS) in which a linear plant is controlled remotely over a network or other shared communication medium. The medium provides a limited number of simultaneous connections, so that only a subset of the plant's sensors and actuators may communicate with the controller at any one time, subject to known transmission delays. Instead of insisting on jointly optimal control and medium access policies, we reduce the infinity of possible access sequences down to those which preserve the stabilisability and detectability of the underlying plant, and are periodic. Our choice of communication and NCS model effect a kind of ‘decoupling’ of the LQG problem, in the sense that the medium access policy can be selected independently of the controller. This guarantees the existence of a stabilising LQG controller which is optimal for the communication policy of choice, and which is then combined with a delay compensator. We include simulations that illustrate our approach.
We discuss controller design for a networked control system (NCS) in which a stochastic linear time invariant (LTI) plant communicates with a controller over a shared medium. The medium supports a limited number of simultaneous connections between the controller and the plant's sensors and actuators, possibly subject to transmission delays. We restrict communication to periodic medium access sequences which preserve the structural properties of the plant, thus decoupling the selection of the communication from that of the controller. Using the plant's controllability/observability indices as a guide for allocating access, we show that the period of the sequences in question can be shorter than previously established. In addition, we explore the use of sequences designed for a simple NCS model, in which sensors and actuators are "ignored" by the controller when they are not actively communicating, in a more complex, but practical, setting that includes zero-order holding. We include a numerical experiment that illustrates our results in the context of LQG control.
The purpose of this paper is to propose a version of causality testing that focuses on how the sign of the returns affects the causality results. We replace the traditional VAR specification used in the Granger causality test by a discrete-time bivariate noisy Mackey glass model. Our test reveals interesting and previously unexplored relationships in US economic series, including inflation, metal, and stock returns.
We discuss stochastic, linear networked control systems (NCSs) in which only a limited number of the plant's sensors and actuators may communicate with the controller at any one time. We explore the problem of designing an LQG controller and an accompanying periodic communication policy, using recent results that forgo optimal communication for the sake of lowering the complexity of the joint (control-communication) problem. We show that the period of the policies in question can be shorter than previously established, and that policies designed under a simpler NCS model, in which sensors and actuators are "ignored" by the controller when they are not actively communicating, can also be effective in the more complex setting which includes zero-order holding (ZOH). Interestingly, the inclusion of a ZOH - although sometimes practical - does not always lead to better performance.
We propose a new server-based efficient protocol for time-release encryption (TRE), or - as sometimes referred to - sending information "into the future". As with other recently-proposed schemes, ours is based on the use of bilinear pairings on any Gap Diffie-Hellman group, allowing absolute release time of the encrypted data. Our protocol possesses the required properties regarding user anonymity and server passivity. It also provides almost-costless scalability in settings with multiple time-servers, and improves significantly upon existing TRE schemes, in terms of computational and communication cost. This makes our approach well-suited to a number of emerging e-applications that require future decryption of confidential data.
We revisit the problem of “sending information into the future” by proposing an anonymous, non-interactive, server-based Timed-Release Encryption (TRE) protocol. We improve upon recent approaches by Blake and Chan, Hwang et al., and Cathalo et al., by reducing the number of bilinear pairings that users must compute, and by enabling additional pre-computations. Our solution compares favorably with existing schemes in terms of computational efficiency, communication cost and memory requirements, and is secure in the random oracle model.
For two parties to communicate securely over an insecure channel, they must be able to authenticate one another and establish a common session key. We propose a new secure one-pass authenticated key establishment protocol which is well suited to one-way communication channels. The protocol is examined using an extension of the Bellare- Rogaway model proposed by Blake-Wilson et. al., and is shown to be provably secure, in the sense that defeating the protocol is equivalent to solving a CDH problem. We compare our protocol to existing approaches, in terms of security and eciency. To the best of our knowledge, ours is the only one-pass protocol that resists general key-compromise imper- sonation attacks, and avoids certain vulnerabilities to loss of information attacks found in other protocols of its class.
We discuss a biologically inspired cooperative control strategy which allows a group of autonomous systems to solve optimal control problems with free final time and partially constrained final state. The proposed strategy, termed “generalized sampled local pursuit” (GSLP), mimics the way in which ants optimize their foraging trails, and guides the group toward an optimal solution, starting from an initial feasible trajectory. Under GSLP, an optimal control problem is solved in many “short” segments, which are constructed by group members interacting locally with lower information, communication and storage requirements compared to when the problem is solved all at once. We include a series of simulations that illustrate our approach.
This paper discusses the stabilization of linear networked control systems (NCSs) in which the medium connecting plant and controller imposes access constraints and known delays. We apply a communication protocol which "decouples" the stabilization problem, in the sense that the medium access policy can be selected independently of the controller while jointly stabilizing the NCS. Our approach is based on reducing the infinity of possible stabilizing communication policies down to a set of sequences which preserve the NCS's controllability and observability. We extend previously established results by treating delays and medium access constraints simultaneously and by demonstrating how a controller designed for a delay-free NCS can be re-used when delays are present. We include a numerical experiment that illustrates our approach
We revisit the problem of "sending information into the future" by proposing an anonymous, non-interactive, server-based Timed-Release Encryp-tion (TRE) protocol. We improve upon recent approaches by Blake and Chan, Hwang et al., and Cathalo et al., by reducing the number of bilinear pairings that users must compute, and by enabling additional pre-computations. Our solution compares favorably with existing schemes in terms of computational efficiency, communication cost and memory requirements, and is secure in the random ora-cle model.
Several ways in which Computer-Aided Control System Design software and rapid prototyping tools have been used to enhance the controls educational program at our university are described. A key to the use in a lecture course has been to provide the students with a computer running the software during every exam. This motivates the students to learn to use the software and facilitates giving exam problems that focus on control design and analysis rather than on mechanical calculations. The use of computer-aided design software and rapid prototyping tools in the lab saves money and allows us to do more complicated experiments and simple projects. It also allows us to emphasize aspects of networked and embedded control systems that are not covered anywhere else within our curriculum.
This paper discusses the stabilization of a networked control system (NCS) in which sensors and actuators of a plant exchange information with a remote controller via a shared communication medium. Access to that medium is governed by a pair of periodic communication sequences. Under the model utilized here, the controller and plant handle communication disruptions by “ignoring” (in a sense to be made precise) sensors and actuators that are not actively communicating. This choice has the effect of significantly reducing the complexity of selecting control/communication policies. It is shown that, for discrete-time NCS, the reachability and observability of the plant can be preserved if the communication sequences are chosen properly. We propose a method for exponentially stabilizing a NCS by first identifying a pair of communication sequences that preserve reachability and observability and then designing an observer-based feedback controller based on those sequences.
Inspired by the process by which ants gradually optimize their foraging trails, this paper investigates the cooperative solution of a class of free final time, partially constrained final state optimal control problems by a group of dynamical systems. We propose an iterative, pursuit-based algorithm which generalizes previously proposed models and converges to an optimal solution by iteratively optimizing an initial feasible trajectory/control pair. The proposed algorithm requires only short-range, limited interactions between group members, avoids the need for a 'global map' of the environment in which the group evolves, and solves an optimal control problem in 'small' pieces, in a manner which will be made precise. The performance of the algorithm is illustrated in a series of simulations and laboratory experiments.
We discuss Kalman filtering and LQ optimal control of a networked control system (NCS) whose sensors and actuators exchange information with a remote controller over a shared communication medium. Access to that medium is governed by a pair of periodic communication sequences. Under the proposed model, the controller and plant handle communication disruptions by "ignoring" sensors and actuators that are not actively communicating. We show that Kalman filtering and LQ optimal control for NCSs can be formulated as a standard LQG problem for an equivalent periodic system. Moreover, under mild conditions, there always exist periodic communication sequences that preserve the detectability and observability of the NCS and thus make it possible to guarantee the existence of a stabilizing LQG controller.
We investigate the solution of a large class of fixed-final-state optimal control problems by a group of cooperating dynamical systems. We present a pursuit-based algorithm, inspired by the foraging behavior of ants that requires each system-member of the group to solve a finite number of optimization problems as it follows other members of the group from a starting to a final state. Our algorithm, termed "sampled local pursuit", is iterative and leads the group to a locally optimal solution, starting from an initial feasible trajectory. The proposed algorithm is broad in its applicability and generalizes previous results. It requires only short-range sensing and limited interactions between group members, and avoids the need for a "global map" of the environment or manifold on which the group evolves. We include simulations that illustrate the performance of our algorithm.
This paper discusses the stabilization of a networked control system (NCS) whose sensors and actuators exchange information with a remote controller over a shared communication medium. Access to that medium is governed by a pair of periodic communication sequences. Under the model utilized here, the controller and plant handle communication disruptions by "ignoring" (in a sense to be made precise) sensors and actuators that are not actively communicating. It is shown that under mild conditions, there exist periodic communication sequences that preserve the reachability and observability of the plant, leading to a straightforward design of a stabilizing feedback controller.
Inspired by the process by which ants gradually optimize their foraging trails, this paper investigates the cooperative solution of a class of free-final time, partially-constrained final state optimal control problems by a group of dynamic systems. A cooperative, pursuit-based algorithm is proposed for finding optimal solutions by iteratively optimizing an initial feasible control. The proposed algorithm requires only short-range, limited interactions between group members, and avoids the need for a "global map" of the environment on which the group evolves. The performance of the algorithm is illustrated in a series of numerical experiments.
Landmark-based graphs are a useful and parsimonious tool for representing large scale environments. Relating landmarks by means of feedback-control algorithms encoded in a motion description language provides a level of abstraction that enables autonomous vehicles to navigate effectively by composing strings in the language to form complex strategies that would be difficult to design at the level of sensors and actuators. In such a setting, feedback control requires one to pay attention not only to sensor and actuator uncertainty, but also to the ambiguity introduced by the fact that many landmarks may look similar when using a modest set of observations. This work discusses the generation of language-based feedback control sequences for landmark-based navigation together with the problem of sensing landmarks sufficiently well to make feedback meaningful. The paper makes two contributions. First, we extend previous work to include the costs of sensing with varying degrees of accuracy. Second, we describe a Monte Carlo based approach to landmark sensing which relies on the use of particle filters. We include simulation results that illustrate our approach.
This work describes a senior-level laboratory/project course at the University of Maryland, College Park (USA), titled "networked and distributed control systems." The course is partially supported by an NSF research and curriculum development grant and brings together undergraduates from three engineering departments (electrical and computer, mechanical, and aerospace) with plans to include chemical, civil, and fire protection engineering. The course combines digital control with networks and information technology. The course goals are to introduce students to emerging areas in systems theory, including topics in networked and distributed control systems; to leverage instructor time and increase productivity in laboratory classes while keeping student teams small; and to minimize costs by using highly adaptable general-purpose equipment.
Preface Part I. Fundamentals Fundamentals of Dynamical Systems Control of Single-Input Single-Output Systems Basics of Sampling and Quantization Discrete-Event Systems Introduction to Hybrid Systems Finite Automata Basics of Computer Architecture Real-Time Scheduling for Embedded Systems Network Fundamentals Part II. Hardware Basics of Data Acquisition and Control Programmable Logic Controllers Digital Signal Processors Microcontrollers SOPCs: Systems on Programmable Chips Part III. Software Fundamentals of RTOS-Based Digital Controller Implementation-Aware Embedded Control Systems From Control Loops to Real-Time Programs Embedded Real-Time Control via MATLAB, Simulink, and xPC Target LabVIEW Real-Time for Networked/Embedded Control Control Loops in RTLinux Part IV. Theory An Introduction to Hybrid Automata An Overview of Hybrid Systems Control Temporal Logic Model Checking Switched Systems Feedback Control with Communication Constraints Networked Control Systems: A Model-Based Approach Control Issues in Systems with Loop Delays Part V. Networking Network Protocols for Networked Control Systems Control Using Feedback over Wireless Ethernet and Bluetooth Bluetooth in Control Embedded Sensor Networks Part VI. Applications Vehicle Applications of Controller Area Network Control of Autonomous Mobile Robots Wireless Control with Bluetooth The Cornell RoboCup Robot Soccer Team: 1999-2003 Index
There is an extensive body of theory and practice devoted to the design of feedback controls for linear time-invariant systems. This chapter contains a brief introduction to the subject with emphasis on the design of digital controllers for continuous-time systems. Before we begin it is important to appreciate the limitations of linearity and of feedback. There are situations where it is best not to use feedback in the control of a system. Typically, this is true for systems that do not undergo much perturbation and for which sensors are either unavailable or too inaccurate. There are also limits to what feedback can accomplish. One of the most important examples is the nonlinearity that is present in virtually all systems due to the saturation of the actuator. Saturation will limit the range of useful feedback gains even when instability does not. It is important to keep this in mind when designing controllers for real systems, which are only linear within a limited range of input amplitudes.
One of systems theory’s most useful and fundamental ideas is that of interconnecting simple systems in order to build complex ones. This is usually accomplished through the use of two important tools. One is a set of theoretical results that help predict the behavior and performance of the composed system given the properties of its components and the manner in which they are connected. The other is the ability to regard the interconnection as ideal in the sense that it neither corrupts nor delays data or—in situations where that is not the case—to “separate” its design from that of the other components (e.g., controllers). The development in recent years of embedded and network technologies has given rise to the area of Networked Control Systems (NCSs), where sensors, actuators and computing elements are connected by means of a network or other shared medium. At the same time, the attempt to expand the scope of systems theory into this new domain has made the assumptions stated above increasingly difficult to justify. The goal of this chapter is to expose some of the complications that arise when a control system includes a network (taken to mean a shared communication medium in the most generic sense) and to introduce a small collection of basic results on the control of systems that operate under communication constraints.
In the last few years, efforts to codify the organizing principles behind biological systems have been capturing the attention of a growing number of researchers in the systems and control community. This endeavour becomes increasingly important as new technologies make it possible to engineer complex cooperating systems, that are nevertheless faced with many of the challenges long overcome by their natural counterparts. One area in particular where biology serves as an inspiring but still distant example, involves systems in which members of a species cooperate to form collectives whose abilities are beyond those of individuals. This paper looks to the process by which ants optimize their foraging trails as inspiration for an organizing principle by which groups of dynamical systems can solve a class of optimal control problems. We explore the use of a strategy termed ‘local pursuit’, which allows members of the group to overcome their limitations with respect to sensing range and available information through the use of neighbour-to-neighbour interactions. Local pursuit enables the group to find an optimal solution by iteratively improving upon an initial feasible control. We show that our proposed strategy subsumes previous pursuit-based models for ant-trail optimization and applies to a large array of problems, including many of the classical situations in optimal control. The performance of our algorithm is illustrated in a series of simulations and experiments.
As modern control theory attempts to elucidate the complexity of systems that combine differential equations and event-driven logic, it must overcome challenges having to do with limited expressive power as well as practical difficulties associated with translating control algorithms into robust and reusable software. The Motion Description Language (MDL) and its "extended" counterpart MDLe, have been at the center of an ongoing effort to make progress on both of these fronts. The goal of this paper is to define MDLe as a formal language, thereby connecting with the vast literature on the subject, and to stimulate experimental work. We discuss the expressive power of MDLe and provide some examples of MDLe programs.
In this work we present an efficient environment representation based on the use of landmarks and language-based motion programs. The approach is targeted towards applications involving expansive, imprecisely known terrain without a single global map. To handle the uncertainty inherent in real-world applications a partially-observed controlled Markov chain structure is used in which the state space is the set of landmarks and the control space is a set of motion programs. Using dynamic programming, we derive an optimal controller to maximize the probability of arriving at a desired landmark after a finite number of steps. A simple simulation is presented to illustrate the approach.
We investigate the simultaneous stabilization of a group of linear systems whose feedback loops are closed over an idealized shared medium. The capacity of that medium is constrained so that only a limited number of controller-plant connections can be accommodated at any one time. We introduce a feedback communication policy - inspired by previous work on queuing systems and real-time scheduling - for deciding which system(s) should be admitted into the network and for how long. The use of feedback in making communication decisions results in a set of autonomous dynamical systems which are coupled to one another due to the presence of communication constraints. We give conditions for the stability of the collection under the proposed communication policy and present simulation results that illustrate our ideas.
This paper considers the use of abstract descriptions of motion control programs and of the environment, and explores some new problems of system theoretic interest that arise as a result. Wc study the problem of active localization for a mobile robot moving on a sparsely-described uncertain environment and show how that problem can be posed as that of observability of a finite automaton. We present algorithms (based on Hidden Markov Models) that answer the question of i) whether or not a representation of the environment (in the form of a directed graph) is observable, and ii) what is the shortest navigation policy that allows the robot to uniquely identify its location on the graph.
One of the important but often overlooked chal-lenges in motion control has to do with the transfer of theoretical tools into software that will allow an autonomous system to interact effectively with the physical world. In a situation familiar to most control practitioners, motion control programs are often machine-specific and are not reusable, even when the underlying algorithm does not require changes. These considerations point out the need for a formal, general-purpose programming lan-guage that would allow one to write motion control programs, incorporating both switching logic and differential equations. The promise held by such a software tool has motivated a research program on the so-called Motion Description Language (MDL) and its extended version MDLe, put forth as device-independent programming languages that can accommodate hybrid controllers, multi-system interactions and agent-to-agent communications. This paper details the syntax, functionality and expressive power of MDLe as well as a software infrastructure that implements the language. We include a set of programming examples that demonstrate the capabilities of MDLe, together with the results of their execution on a group of mobile robots.
Recently developed experimental and numerical environments have helped breathe life into the various control theories found in textbooks and have thereby greatly changed the educational experience of students of automatic control. Nonlinear balance beams, inverted pendulums, and distributed parameter thermal systems are now widely available for hands-on experimentation. Many students react quite positively to this additional dose of realism. Because the models selected for such experiments are usually accurately described by relatively simple differential equations, the laboratory experience reinforces both the textbook analysis and the value of numerical simulation. At the same time, there is a growing realization among educators and employers that students of automatic control should be encouraged to think of the subject in broader terms. The systems approach should embrace communica-
We propose a landmark-based representation of maps to be used for robot navigation and exploration. Our approach is aimed towards mobile robots that operate over expansive, imprecisely known terrain without a single "global" map. Instead, a map is pieced together from local terrain and navigation data stored in a directed graph. Each of the graph's vertices contains information describing a landmark locally (e.g. a detailed map of that landmark's immediate surroundings). The geometric relationships between landmarks are unknown. Graph edges store language-based directions that enable a robot to steer between landmarks. These directions are written in the motion description language MDLe, reducing the complexity of the map and making navigation programs robot-independent. Furthermore, the proposed architecture is economical with respect to the amount of storage required to describe far-flung areas of interest. We present preliminary results demonstrating our ideas using an indoor robot
We study the dynamics and explore the controllability of a family of sphere-plate mechanical systems. These are nonholonomic systems with a five-dimensional (5-D) configuration space and three independent velocities. They consist of a sphere rolling in contact with two horizontal plates. Kinematic models of sphere-plate systems have played an important role in the control systems literature addressing the kinematics of rolling bodies, as well as in discussions of nonholonomic systems. However, kinematic analysis falls short of allowing one to understand the dynamic behavior of such systems. We formulate and study a dynamic model for a class of sphere-plate systems in order to answer the question: -is it possible to impart a net angular momentum to a sphere which rolls without slipping between two plates, given that the position of the top plate is subject to exogenous forces?”
Control and communication issues are traditionally “decoupled” in discussions of decision and control problems, as this simplifies the analysis and generally works well for classical models. This fundamental assumption deserves re-examination as control applications spread into new areas where system complexity is significant. Such areas include the coordinated control of aerial vehicles (UAVs), MEMS devices, multi-joint manipulators and other settings where many systems must share the attention of a decision-maker. We consider a new class of sampled-data systems (termed “computer-controlled systems”) that offer the possibility of jointly optimizing between control and communication goals. Computer-controlled LTI systems can be viewed as linear operators between appropriate inner-product spaces. The generalized inverses of these operators are used to solve a class of finite-horizon tracking problems
In this paper we will discuss the control of a five-fingered, 18 degree of freedom, tendon driven robot hand known as the Atlas Anthrobot. Developing a traditional PID controller for this hand has proved difficult because the behavior of the tendon drive system produces is highly nonlinear. This nonlinear nature of the system however does suggest it as a candidate for fuzzy control. Along these lines we have developed a PID control system and a fuzzy control system for the hand which are both capable of simultaneously controlling the position of the fingertips and the amount of force exerted through a contact point on the fingertip. In this paper we will present a comparison between the two systems in order to generally make statements about the appropriateness of each controller for controlling complicated tendon driven devices such as our robot hand
We propose a new server-based protocol for timed- release encryption (TRE), sometimes referred to as "sending infor- mation into the future". As with other recently-proposed schemes, ours is based on the use of bilinear pairings on a Gap Diffie-Hellman group, and allows a sender to specify with precision the release time of the encrypted data. Our protocol possesses various required prop- erties related to security, user anonymity and server passivity. It also provides almost-costless scalability in settings with multiple time- servers, and improves significantly upon existing TRE schemes, in terms of communication cost. The basic version of our protocol is extended to include new desirable features, such as message pre- opening, and "hiding" of important details associated with a cipher- text, thus making our approach well-suited to a number of emerging e-applications that require future decryption of confidential data.
This paper explores the use of a bio-inspired control algorithm, termed "local pursuit", as a numerical tool for computing optimal control-trajectory pairs in settings where analytical solutions are difficult to obtain. Inspired by the foraging activities of ant colonies, local pursuit has been the focus of recent work on cooperative optimization. It allows a group of agents to solve a broad class of optimal control problems (including fixed final time, partially-constrained final state problems) and affords certain benefits with respect to the amount of information (description of environment, coordinate systems, etc.) required to solve the problem. Here, we present a numerical optimization method that combines local pursuit with the well-known technique of multiple shooting, and compare the computational efficiency and capabilities of the two approaches. The proposed method can overcome some important limitations of multiple shooting by solving an optimal control problem "in small pieces". Specifically, the use of local pursuit increases the size of the problem that can be handled under a fixed set of computational resources. Furthermore, local pursuit can be effective in some situations where multiple shooting leads to an ill-conditioned nonlinear programming problem. The trade-off is an increase in computation time. We compare our pursuit-based method with direct multiple shooting using an example that involves optimal orbit transfer of a simple satellite.
Cooperative control systems are increasingly emerging as significant alternatives to their centralized counterparts. The rising interest in deploying cooperative systems is fueled by the development of decentralized systems with low cost and performance advantages. For example, mobile exploration and information gathering tasks can often be accomplished cheaply and more reliably by swarms of small autonomous robots as opposed to a single more sophisticated one. Cooperative control is also applied in many tasks that can not be performed by a single system, e.g. satellite arrays that enable global communication, geographically remote systems that communicate via network and others. The goal of our research is to investigate optimal control in cooperative systems, using algorithms inspired from biology. We begin with a review of collective behavior in biological systems.
develops, applies and teaches advanced methodologies of design and analysis to solve complex, hierarchical, heterogeneous and dynamic problems of engineering technology and systems for industry and government. Abstract Inspired by the process by which ants gradually optimize their foraging trails, this paper investigates the cooperative solution of a class of free-final time, partially-constrained final state optimal control problems by a group of dynamical systems. We propose a cooperative, pursuit-based algorithm which generalizes previously-proposed models and converges to an optimal solution by iteratively optimizing an initial feasible trajectory/control pair. The proposed algorithm requires only short-range, limited interactions between group members, avoids the need for a "global map" of the environment in which the group evolves, and solves an optimal control problem in "small" pieces, in a manner which will be made precise. The performance of the algorithm is illustrated in a series of simulations and laboratory experiments.
We investigate the solution of a large class of fixed-final-state optimal control problems by a group of cooperating dynamical systems. We present a pursuit-based algorithm inspired by the foraging behavior of ants that requires each system-member of the group to solve a finite number of optimization problems as it follows other members of the group from a starting to a final state. Our algorithm, term "sampled local pursuit", is iterative and leads the group to a locally optimal solution, starting from an initial feasible trajectory. The proposed algorithm is broad in its applicability and generalizes previous results; it requires only short-range sensing and limited interactions between group members, and avoids the need for a "global map" of the environment or manifold on which the group evolves. We include simulations that illustrate the performance of our algorithm.

Citations

Affiliations

University of Macedonia
Department
  • Department of Applied Informatics
University of Maryland, College Park
Department
  • Department of Electrical & Computer Engineering
  • Department of Mechanical Engineering
Harvard University
Department
  • School of Engineering and Applied Sciences

Disciplines

Linguistics
Topic

Publication Stats

Citations
1,449
Join ResearchGate to find the people and research you need to help your work.
  • 15+ million members
  • 118+ million publications
  • 700k+ research projects
Join for free