Person:

Egerstedt, Magnus B.

Permanent Link

https://hdl.handle.net/1853/72204

Associated Organization(s)

Organizational Unit

School of Electrical and Computer Engineering

Full item page

Publication Search Results

Now showing 1 - 10 of 49

Barrier Functions and Model Free Safety With Applications to Fixed Wing Collision Avoidance

(Georgia Institute of Technology, 2021-07-29) Squires, Eric G.

Robotics is now being applied to a diversity of real-world applications and in many areas such as industrial, medical, and mobile robotics, safety is a critical consideration for continued adoption. In this thesis we therefore investigate how to develop algorithms that improve the safety of autonomous systems using both a model-based and model-free framework. To begin, we make a variety of assumptions (e.g., that a model is known, there is a single safety constraint, there are no communication limits, and that the state can be sensed everywhere), and show how to guarantee the safety of the system. The contribution of the initial approach is a generalization of an existing method for creating a barrier function, which is a function similar to a Lyapunov function that can be used to make safety guarantees. We then investigate relaxing these initial assumptions. In some cases, new additional assumptions are required, performance may be reduced, or safety guarantees may no longer be available. We motivate the thesis with collision avoidance for fixed wing aircraft which can be viewed as a pairwise constraint on each pair of aircraft. This introduces the need for considering multiple safety factors simultaneously, and we show that an additional assumption is needed in this case. We then relax the assumption that the vehicles have unlimited communication and find that safety can still be guaranteed. However, it is possible in this case that the overriding safety controller may be more invasive than if more communication is allowed. When we then further relax the assumption that the state can be sensed at all times, safety can still be guaranteed in some specified situations but the system may be more permissive in approaching safety boundaries. We finally remove the assumption of a known model for dynamics. Although removing this assumption means the system is no longer guaranteed to be safe, the benefit is that it allows a safety designer to build a far less invasive override to get more performance out of the system.
Collective behavior and task persistification in lazy and minimalist collectives

(Georgia Institute of Technology, 2021-05-10) Dutta, Bahnisikha

When individuals in a collective system are constrained in terms of sensing, memory, computation, or power reserves; the design of algorithms to control them becomes challenging. These individual limitations can be due to multiple reasons like the shrinking size of each agent for bulk manufacturing efficiency or enforced simplicity to attain cost efficiency. Whereas, in some areas like nano-medicine, the nature of the task itself warrants such simplicity. This thesis presents algorithms inspired by biological and statistical physics models to achieve useful collective behavior through simple local physical interactions and, minimalist approaches to persistify tasks for long durations in collectives with limited capabilities and energy reserves. The first part of the thesis presents a system of vibration-driven robots that embodies the features of simplicity described above. A combination of theory, experiment, and simulation is used to study dynamic aggregation behavior in these robots facilitated via short-range physical attraction potentials between agents. Collectives in a dynamically aggregated state are shown to be capable of transporting objects over relatively long distances in a finite arena. In the rest of the thesis, two different, yet complementary systems are studied and elaborated to highlight the usefulness of distributed inactivity and activity modulation in aiding persistification of tasks in collectives incapable of implementing complicated algorithms to incorporate regular energy replenishing cycles. To summarize, an approach to achieving dynamic aggregation and related tasks like object transport in a constrained brushbot system is described. Two different artificial and biological collective systems are explored to reveal strategies through which tasks can be persistified without requiring complicated computations, sensing, and memory.
Coverage control: From heterogeneous robot teams to expressive swarms

(Georgia Institute of Technology, 2020-07-27) Santos Fernandez, Maria Teresa

Coverage control constitutes a canonical multi-robot coordination strategy that allows a collection of robots to distribute themselves over a domain to optimally monitor the relevant features of the environment. This thesis examines two different aspects of the coverage problem. On the one hand, we investigate how coverage should be performed by a multi-robot team with heterogeneous sensor equipment in the presence of qualitatively different types of events or features in the domain, which may evolve over time. To this end, different information exchange strategies among the robots are considered, and the performance of the resulting distributed control laws is compared experimentally on a team of mobile robots. In addition, we present a constraint-based approach that allows the multi-robot team to cover different types of features whose locations in the domain may evolve other time. On the other hand, in the context of swarm robotics in the arts, this thesis investigates how the coverage paradigm, which affords the control of the entire multi-robot team through the high-level specification of density functions, can serve as an effective interaction modality for artists to effectively utilize robotic swarms in different forms of art expression. In particular, we explore the use of coverage, along with other standard multi-robot control algorithms, to create emotionally expressive behaviors for robot theatre applications. Furthermore, the heterogeneous coverage framework developed in this thesis is employed to interactively control desired concentrations of color throughout a canvas for the purpose of artistic multi-robot painting.
Heterogeneous interaction modalities for shape-similar formations

(Georgia Institute of Technology, 2020-03-18) Buckley, Ian Howell

Formation control of multi-robot teams is fundamentally influenced by the available sensing and communication capabilities of individual robots. The significance of these capabilities manifests in the network topology induced by interaction modalities present in the team, which may include maintenance of relative distances, bearings, or angles in the formation. To understand this significance and aid in design of effective control strategies, this thesis investigates the interplay between network topology and heterogeneous interaction modalities present in multi-robot formations. With regard to this investigation, each chapter of this thesis addresses a series of research questions that motivate and drive the results. The thesis begins by considering formations in which the relative angles between robots are maintained. To characterize such formations, infinitesimal shape-similarity is developed to describe frameworks in which angle maintenance renders the framework invariant to infinitesimal translations, rotations, and uniform scaling. After developing tools for assessing frameworks for this property, design of formation controllers for infinitesimally shape-similar frameworks reveals the sensing and communication requirements on the robots executing them. To explore relaxations of these requirements, a bearing-only self-assembly mechanism for a class of infinitesimally shape-similar frameworks is designed, and a formation-control strategy is developed to leverage a single distance measurement, suggesting that heterogeneity may be exploited at large. To relate heterogeneous distance, bearing, and angle constraints, the relationships between infinitesimal rigidity, bearing-rigidity, and shape-similarity are examined, espousing the coupling of network topology and interaction modalities in a team. The motions of formations specified by heterogeneous constraints are then characterized, and formation-control strategies are developed. Ultimately, this thesis demonstrates that the coupling of the network topology and heterogeneous interaction modalities of multi-robot teams should be accounted for explicitly to assess the tradeoffs between connectivity and information access in achieving effective formation control.
Local encounters in robot swarms: From localization to density regulation

(Georgia Institute of Technology, 2019-11-11) Mayya, Siddharth

In naturally occurring swarms---living as well as non-living---local proximity encounters among individuals or particles in the collective facilitate a broad range of emergent phenomena. In the context of robot swarms operating with limited sensing and communication capabilities, this thesis demonstrates how the systematic analysis of inter-robot encounters can enable the swarm to perform useful functions without the presence of a central coordinator. We combine ideas from stochastic geometry, statistical mechanics, and biology to develop mathematical models which characterize the nature and frequency of inter-robot encounters occurring in a robot swarm. These models allow the swarm to perform functions like localization, task allocation, and density regulation, while only requiring individual robots to measure the presence of other robots in the immediate vicinity---either via contact sensors or binary proximity detectors. Moreover, the resulting encounter-based algorithms require no communication among the robots or the presence of a central coordinator, and are robust to individual robot failures occurring in the swarm. Throughout the thesis, experiments conducted on real robot swarms vindicate the idea that inter-robot encounters can be advantageously leveraged by individuals in the swarm.
Specification composition and controller synthesis for robotic systems

(Georgia Institute of Technology, 2019-03-21) Glotfelter, Paul

From precision agriculture to autonomous-transportation systems, robotic systems have been proposed to accomplish a number of tasks. However, these systems typically require satisfaction of multiple constraints, such as safety or connectivity maintenance, while completing their primary objectives. The objective of this thesis is to endow robotic systems with a Boolean-composition and controller-synthesis framework for specifications of objectives and constraints. Barrier functions represent one method to enforce such constraints via forward set invariance, and Lyapunov functions offer a similar guarantee for set stability. This thesis focuses on building a system of Boolean logic for barrier and Lyapunov functions by using min and max operators. As these objects inherently introduce nonsmoothness, this thesis extends the theory on barrier functions to nonsmooth barrier functions and, subsequently, to controlled systems via control nonsmooth barrier functions. However, synthesizing controllers with respect to a nonsmooth function may create discontinuities; as such, this thesis develops a controller-synthesis framework that, despite creating discontinuities, still produces valid controllers (i.e., ones that satisfy the objectives and constraints). These developments have been successfully applied to a variety of robotic systems, including remotely accessible testbeds, autonomous-transportation scenarios, and leader-follower systems.
Power-aware hybrid-dynamical approach to coverage control in multi-robot systems

(Georgia Institute of Technology, 2018-04-24) Olsen, Mark Ryan

This thesis develops an algorithm which allows robots in a multi-robot team to optimize for battery power while performing coverage control so as to maximize the mission life of the multi-robot team. We envision a scenario where robots with limited battery supply are executing the well known Lloyd's algorithm in order to effectively cover a certain region. We perform a trade-off between the distance of a robot from the centroid of its Voronoi cell, and the energy required to traverse that distance. In order to execute this trade-off two different strategies are presented -- in one case, the reduction in cost due to coverage is compared against the energy required to traverse the distance to the centroid, and using a user-defined threshold, the decision is made. Then, a more sophisticated algorithm is used to perform the trade-off where the robots solves a switch-time optimization problem to decide whether it should move or it should stay.
A control theoretic perspective on social networks

(Georgia Institute of Technology, 2018-04-05) Ruf, Sebastian Felix

This thesis discusses the application of control theory to the study of complex networks, drawing inspiration from the behavior of social networks. There are three topic areas covered by the thesis. The first area considers the ability to control a dynamical system which evolves over a network. Specifically, this thesis introduces a network controllability notion known as herdability. Herdability quantifies the ability to encourage general behavioral change in a system via a set-based reachability condition, which describes a class of desirable behaviors for the application of control in a social network setting. The notion is closely related to the classical notion of controllability, however ensuring complete controllability of large complex networks is often unnecessary for certain beneficial behaviors to be achieved. The basic theory of herdability is developed in this thesis. The second area of study, which builds directly on the first, is the application of herdability to the study of complex networks. Specifically, this thesis explores how to make a network herdable, an extension of the input selection problem which is often discussed in the context of controllability. The input selection problem in this case considers which nodes to select to ensure the maximal number of nodes in the system are herdable. When there are multiple single node sets which can be used to make a system completely herdable, a herdability centrality measure is introduced to differentiate between them. The herdability centrality measure, a measure of importance with respect to the ability to herd the network with minimum energy, is compared to existing centrality measures. The third area explores modeling the spread of the adoption of a beneficial behavior or an idea, in which the spread is encouraged by the action of a social network. A novel model of awareness-coupled epidemic spread is introduced, where agents in a network are aware of a virus (here representing something which should be spread) moving through the network. If the agents have a high opinion of the virus, they are more likely to adopt it. The behavior of this viral model is considered both analytically and in simulation.
Multi-robot coordination and safe learning using barrier certificates

(Georgia Institute of Technology, 2018-04-03) Wang, Li

The objective of this research is to develop a formal safety framework for collision-free and connectivity sustained motion in multi-robot coordination and learning based control. This safety framework is designed with barrier certificates, which provably guarantee the safety of dynamical systems based on the set invariance principle. The barrier certificates are enforced on the system using an online optimization-based controller such that minimal changes to the existing control strategies are required to guarantee safety. The proposed safety barrier certificates are validated on real multi-robot systems consisting of multiple Khepera robots, Magellan Pro robot, GRITS-Bots, and Crazyflie quadrotors.
Mixed centralized/decentralized coordination protocols for multi-agent systems

(Georgia Institute of Technology, 2017-04-10) Hale, Matthew Thomas

This thesis uses a mixture of centralized and decentralized architectures and algorithms to develop coordination strategies for multi-agent systems. Conventionally, centralized and decentralized methods are viewed as belonging to distinct paradigms, each with its own features and drawbacks, and multi-agent coordination algorithms are typically classified as being exclusively one or the other. However, emerging technologies such as cloud computing make it feasible to incorporate some centralization into an otherwise decentralized system, and one may ask how to embrace this mix of centralized and decentralized information that is rapidly being integrated into various systems such as the smart power grid, swarms of robots, and cyber-physical systems. To address this question, two problem domains are considered. The first is that of asynchronous coordination, in which agents generate and share information with arbitrary timing. The second concerns private coordination, in which teams of agents must work together without revealing sensitive information. In both cases, mixing centralized and decentralized information enables successful coordination despite the challenges imposed by asynchrony and privacy, and theoretical performance guarantees are derived for each algorithm that is developed. Complementing these theoretical developments, robotic experiments are included that demonstrate the utility of these algorithms in practice.