IEEE.org | IEEE Xplore Digital Library | IEEE Standards | IEEE Spectrum | More Sites
Call for Award Nominations
More Info
Urban Air Mobility (UAM) is an emerging aviation sector and is playing an integral part in the on-demand mobility revolution. UAM is powered by the convergence of advances in distributed electrical propulsion (DEP) and vehicle autonomy. The complexity of operations in the urban environment and the unconventional vehicle configurations designed to take advantage of new propulsion technologies, result in numerous challenges that benefit from a control-centric approach. In this talk, we outline some of these challenges and present our current approach to addressing them. For example, in order to achieve full market potential and access to UAM, vehicle autonomous flight is required. A key barrier to autonomous flight in a large multi-agent system is dealing with off-nominal situations and contingencies in a safe and predictable manner. We present our approach to intelligent contingency management and share recent results and open problems. Additionally, we discuss another major barrier to ubiquitous UAM – the noise signature produced by vehicles with multiple rotors. We present our approach to minimizing such noise within the framework of the acoustically-aware vehicle.
In this talk, we will discuss how optimization and control theory play a fundamental, and often overlooked, role in multi-UAV coordination. We will see how the solutions of optimal control problems are essential in combinatorial assignment algorithms. Using intuition gained by solving these problems, one can intuit how results dealing with static task assignments extend to cases where the tasks are dynamic in nature. The concepts discussed in this talk will be highlighted with specific problems that are relevant to defense applications.
Genetic circuits control every aspect of life and thus the ability to engineer them de-novo opens exciting possibilities, from revolutionary drugs and green energy to bugs that recognize and kill cancer cells. The robustness of natural gene networks is the result of a million years of evolution and is in contrast with the fragility of today’s engineered circuits. A genetic module’s input/output behavior changes in unpredictable ways upon inclusion into a larger system. Therefore, each component of a system is usually redesigned every time a new piece is added. Rather than relying on such ad-hoc design procedures, control theoretic approaches may be used to engineer “insulation” of circuit components from context, thus enabling modular composition through specified input/output connections. In this talk, I will give an overview of modularity failures in genetic circuits, focusing on problems of loads, and introduce a control-theoretic framework, founded on the concept of retroactivity, to address the insulation question. Within this framework, insulation can be mathematically formulated as a disturbance rejection problem; however, classical solutions are not directly applicable due to biophysical constraints. I will thus introduce solutions relying on time-scale separation, a key feature of biomolecular systems, which were used to build two devices: the load driver and the resource decoupler. These devices aid modularity, facilitate predictable composition of genetic circuits and show that control-theoretic approaches may be suitable to address pressing challenges in engineering biology.
Recent results in deep learning have left no doubt that it is amongst the most powerful modeling tools that we possess. The real question is how can we utilize deep learning for control without losing stability and performance guarantees. Even though recent successes in deep reinforcement learning (DRL) have shown that deep learning can be a powerful value function approximator, several key questions must be answered before deep learning enables a new frontier in robotics. DRL methods have proven difficult to apply to real-world robotic systems where stability matters and safety is critical. In this talk, I will present our recent work in bringing deep learning-based methods to provably stable adaptive control and expand upon possibilities of using concepts from adaptive control to create safe and stable reinforcement learning algorithms. I will put our theoretical work in context by discussing several applications in flight control and agricultural robotics. I will also bring to light our recent work in understanding how the octopus brain works and how it can inspire future learning and distributed control tools.
Swarm robotics, a subfield of both robotics and artificial swarm intelligence, focuses on the development of teams composed of large numbers of autonomous robotic agents. Like swarm intelligence, swarm robotics arises from the study of the phenomenology of biological systems in which large numbers of individuals collaborate in joint collective actions for the benefit of the community as a whole. However, whereas swarm intelligence often utilizes the means and mechanisms of bio-inspired swarms for numerical optimization, the goals of bio-inspired robot swarms are generally concerned with the use of large numbers of low-cost physically embodied agents, acting together in a real-world environment, to achieve a common purpose. This talk will discuss key methods and bio-inspired algorithms for use in programming and controlling robotic swarms, and potential applications of these swarms.
Reachability analysis, which considers computing or approximating the set of future states attainable by a dynamical system over a time horizon, is receiving increased attention motivated by new challenges in, e.g., learning-enabled systems, assured and safe autonomy, and formal methods in control systems. Such challenges require new approaches that scale well with system size, accommodate uncertainties, and can be computed efficiently for in-the-loop or frequent computation. In this talk, we present and demonstrate a suite of tools for efficiently over-approximating reachable sets of nonlinear systems based on the theory of mixed monotone dynamical systems. A system is mixed monotone if its vector field or update map is decomposable into an increasing component and a decreasing component. This decomposition allows for constructing an embedding system with twice the states such that a single trajectory of the embedding system provides hyperrectangular over-approximations of reachable sets for the original dynamics. This efficiency can be harnessed, for example, to compute finite abstractions for tractable formal control verification and synthesis or to embed reachable set computations in the control loop for runtime safety assurance. We demonstrate these ideas on several examples, including an application to safe quadrotor flight that combines runtime reachable set computations with control barrier functions implemented on embedded hardware.
The fields of adaptive control and machine learning have evolved in parallel over the past few decades, with a significant overlap in goals, problem statements, and tools. Machine learning as a field has focused on computer-based systems that improve and learn through experience. Oftentimes the process of learning is encapsulated in the form of a parameterized model such as a neural network, whose weights are trained in order to approximate a function. The field of adaptive control, on the other hand, has focused on the process of controlling engineering systems in order to accomplish regulation and tracking of critical variables of interest. Learning is embedded in this process via online estimation of the underlying parameters. In comparison to machine learning, adaptive control often focuses on limited-data problems where fast, on-line performance is critical. Whether in machine learning or adaptive control, this learning occurs through the use of input-output data. In both cases, the approach used for updating the parameters is often based on gradient descent-like and other iterative algorithms. Related tools of analysis, convergence, and robustness in both fields have a tremendous amount of similarity. As the scope of problems in both topics increases, the associated complexity and challenges increase as well. In order to address learning and decision-making in real time, it is essential to understand these similarities and connections to develop new methods, tools, and algorithms.
This talk will examine the similarities and interconnections between adaptive control and optimization methods commonly employed in machine learning. Concepts in stability, performance, and learning, common to both fields will be discussed. Building on the similarities in update laws and common concepts, new intersections and opportunities for improved algorithm analysis will be explored. High-order tuners and time-varying learning rates have been employed in adaptive control leading to very interesting results in dynamic systems with delays. We will explore how these methods can be leveraged to lead to provably correct methods for learning in real-time with guaranteed fast convergence. Examples will be drawn from a range of engineering applications.
Hiring and Supporting a Diverse Faculty (Dr. Bonnie Ferri)
This talk will explore some of the issues, challenges, and opportunities for hiring and supporting a diverse faculty in STEM disciplines. What are some policies, practices, and programs that support a healthy and productive culture among a diverse population? Do our promotion and advancement practices need retuning? What contributions can a professional society have to support success? Finally, what can each of us do individually to support diversity, equity, and inclusion in the faculty ranks?
Logistics and transportation systems are man-made systems that are well suited for modeling in a discrete event system framework and particularly by Petri Nets (PNs), due to their different characteristics: distributed, parallel, deterministic, stochastic, discrete, and continuous. The paper presents a survey on the various Petri nets modeling frameworks proposed in the related literature for logistics and transportation systems, with applications to modeling, simulation, analysis, optimization and control. In particular, we focus on papers dealing with freight transportation and outline and classify the related works conducted using PNs as regards the proposed framework and addressed problems. We also debate the approach's viability, discussing contributions and limitations, and identify future research potentials.
I will give a brief tutorial of synthesis in a distributed setting, where the goal is to automatically synthesize, if they exist, a set of distributed observers or controllers which together achieve a specific goal. Rather than giving an exhaustive survey I will focus on specific aspects of the problem. In particular, my talk will be structured in two parts. In the first part I discuss theoretical aspects, specifically decidability and undecidability. In the second part I discuss distributed synthesis in practice, specifically, automatically synthesizing protocols such as the Alternating Bit Protocol using a combination of example scenarios and formal specifications as user inputs.
In this talk, we will give an overview of finite abstractions, which are graph-based representations for continuous-state control systems. If these finite abstractions are constructed properly, they can be used to design controllers using techniques from discrete event systems or reactive synthesis in a way that the designed controller can be implemented on the underlying continuous control system (namely, the concrete system) and provide guarantees on the closed-loop behavior. In order to lead to a correct-by-construction design, the abstract system should satisfy a certain relation with the concrete system. We will introduce several such relations including, (bi)simulation relations, over-approximations, feedback refinement relations, and discuss what type of properties are preserved under these relations. Finally, we will discuss various ways of constructing these abstractions, e.g., based on gridding or partitioning the state space, for different classes of systems, e.g., discrete-time or continuous-time. Several examples will be used throughout to demonstrate these techniques in action. The talk will conclude with a summary of more recent results and a discussion on several research directions.
We discuss recursive algorithms for state estimation and event inference, both of which are key tasks for monitoring and control of discrete event systems. In particular, we discuss algorithms for current-, initial-, and delayed-state estimation. We also discuss implications to various pertinent properties of interest, such as detectability (i.e., the ability to determine the exact system state after a finite number of events), diagnosability (i.e., the ability to detect within finite time the occurrence/type of a fault), and opacity (i.e., the guarantee that outsiders will never be able to infer that the system state lies within a set of certain secret/critical states). The talk also briefly discusses the extension of state estimation and event inference methodologies in emerging decentralized/distributed observation settings.
Opacity is an information-flow property used in privacy and security applications. A dynamic system is opaque if an external observer that knows the system model and makes online observations of its behavior is not able to detect with certainty some "secret" information about the system. We discuss various notions of opacity and their verification in the context of discrete event systems modeled by automata or transition systems: current-state opacity, initial-state opacity, and K-step opacity. Then we consider how to enforce opacity for systems that are not opaque. We focus on opacity enforcement using obfuscation, when an external interface edits the outputs of the system in order to confuse the observer. We present solution methodologies for different variations of this problem. We conclude with illustrative examples of opacity in the context of location privacy in location-based services.
Snake robots are motivated by the slender and flexible body of biological snakes, which allows them to move in virtually any environment on land and in water. Since the snake robot is essentially a manipulator arm that can move by itself, it has a number of interesting applications including firefighting and search-and-rescue operations. In water, the robot is a highly flexible and dexterous manipulator arm that can swim by itself like a sea snake. This highly flexible snake-like mechanism has excellent accessibility properties, and not only can the snake robot access narrow openings and confined areas, it can also carry out highly complex manipulation tasks at this location since manipulation is an inherent capability of the system. This talk presents research results on modelling, analysis and control of snake robots, including both theoretical and experimental results. Ongoing efforts are described for bringing the results from university research towards industrial use.
Machine learning-based techniques have recently revolutionized nearly every aspect of autonomy. In particular, deep reinforcement learning (RL) has rapidly become a powerful alternative to classical model-based approaches to decision-making, planning, and control. Despite the well-publicized successes of deep RL, its adoption in complex and/or safety-critical tasks at scale and in real-world settings is hindered by several key issues, including high sample complexity in large-scale problems, limited transferability, and lack of robustness guarantees. This talk explores our recently developed solutions that address these fundamental challenges for both single and multiagent RL. In addition, this talk highlights the complementary role that classical model-based techniques can play in synergy with data-driven methods in overcoming these issues. Real experiments with ground and aerial robots will be used to illustrate the effectiveness of the proposed techniques. The talk will conclude with an assessment of the state of the art and highlight important avenues for future research.
There are many interesting dynamical systems that can be regarded as hierarchically networked systems in a variety of fields including control. One of the ideas to treat those systems properly is "Glocal (Global/Local) Control," which means that the global purpose is achieved by local actions of measurement and control cooperatively. The key for realization of glocal control is hierarchically networked dynamical systems with multiple resolutions in time and space depending on the layer, which introduce many new theoretical control challenges aiming at practical effectiveness in synthetic biology and engineering. The main issues may include how to achieve synchronization by decentralized control and how to make a compromise of two different objectives, one for global and the other for local operations. The background, the idea, and the concept of glocal control are addressed based on an understanding of Internet of Things (IoT) from the control perspective. This talk presents two research topics, namely, (1) hierarchically decentralized control for networked dynamical systems, and (2) robust instability analysis for a class of uncertain nonlinear networked systems.
Regarding the first topic, we propose a theoretical framework for hierarchically decentralized control of networked dynamical systems that can take account of the tradeoff between the global and local objectives to achieve the desired harmony under change of the environments. Several new ideas, by exploiting the special structure of the target systems, enable us to develop scalable control design methods based on the powerful theory in classical, modern, and robust control. The effectiveness of the new theoretical foundations on the analysis and synthesis is experimentally confirmed by applications to electric vehicle control.
The second topic is quite new. It is on robust instability analysis for guaranteed persistence of nonlinear oscillations in the presence of a dynamic perturbation, which is important in synthetic biology. The problem of robust instability has a very different feature from that of robust stability, and hence a new theoretical setting is needed. We define the instability margin as the infimum of the H-infinity norm of the stable perturbation that stabilizes an equilibrium point for a class of nonlinear networked systems. To this end, we introduce a notion of the robust instability radius (RIR) for linear systems and provide a systematic way of finding the exact RIR. Based on this result, the instability margin can be analyzed exactly, with an additional theoretical investigation on how to properly treat the change of the equilibrium point due to the perturbation. The results are applied to the Repressilator in synthetic biology, and the effectiveness is confirmed by numerical simulations.
Mathematics plays a fundamental role in disciplines such as physics, engineering, computer science, and chemistry and has been more recently accepted as a suitable language for solving problems in biology, biochemistry, and medicine.
Control theory is part of the mathematical world and has the peculiarity of borrowing tools from different branches of mathematics. Interestingly, many of the techniques conceived and routinely used to solve control problems can be quite successfully adapted to solve new relevant problems, both practical and curiosity-driven, in other fields.
This talk discusses the structural analysis of systems, aimed at explaining how mechanisms work, why they work in a certain way, and to which extent they perform their task properly even in the presence of perturbations and disturbances.
The first part of the talk briefly introduces some preliminary motivating examples of mechanisms, borrowed from other disciplines alien to control theory, to show how a control approach can be very powerful to understand fundamental principles.
The second part introduces the definitions of structural versus robust properties, discussing paradigmatic case studies from the literature. Robust stability analysis is presented in an inverse form: "We know that this system is stable, but why is the system so incredibly stable?". Other fundamental concepts such as (perfect) adaptation, structural steady-state analysis, graph loop analysis, and aggregation are considered.
The third part discusses application examples from biology and biochemistry, to showcase the potential impact that the mathematical approach of control theory, suitably revised, can have in these disciplines and how interdisciplinary research can bring fresh ideas to control theorists.
Existing control design and verification methods are limited in their ability to address large numbers of interacting agents, multiple layers of feedback, and complex system-level requirements. This talk will demonstrate a strategy for overcoming this limitation with compositional and hierarchical approaches. The compositional approach exposes a complex system as an interconnection of smaller subsystems and derives system-level guarantees from subsystem properties. The hierarchical approach decomposes the synthesis and verification tasks into layers, from high-level decision making to low-level control synthesis. Taken together, these approaches break apart intractably large design and verification problems into subproblems of manageable size. In addition to broadly applicable methodology, the talk will present numerous motivating applications and experimental results, involving multicellular biological systems, fleets of autonomous vehicles, and a multiscale traffic management system.
Control policies that involve the real-time solution of one or more convex optimization problems include model predictive (or receding horizon) control, approximate dynamic programming, and optimization-based actuator allocation systems. They have been widely used in applications with slower dynamics, such as chemical process control, supply chain systems, and quantitative trading, and are now starting to appear in systems with faster dynamics. In this talk I will describe a number of advances over the last decade or so that make such policies easier to design, tune, and deploy. We describe solution algorithms that are extremely robust, even in some cases division free, and code generation systems that transform a problem description expressed in a high-level domain-specific language into source code for a real-time solver suitable for control. The recent development of systems for automatically differentiating through a convex optimization problem can be used to efficiently tune or design control policies that include embedded convex optimization.
More information provided here.
