In this thesis, we consider a swarm of military drones flying over an unfriendly territory, where a drone can be shot down by an enemy with an age-based risk probability. We study the problem of scheduling surveillance image transmissions among the drones with the objective of minimizing the overall casualty. We present Hector, a reinforcement learning-based scheduling algorithm. Specifically, Hector only uses the age of each detected target, a piece of locally available information at each drone, as an input to a neural network to make scheduling decisions. Extensive simulations show that Hector significantly reduces casualties than a baseline round-robin algorithm. Further, Hector can offer comparable performance to a high-performing greedy scheduler, which assumes complete knowledge of global information. / Master of Science / Drones have been successfully deployed by the military. The advancement of machine learning further empowers drones to automatically identify, recognize, and even eliminate adversary targets on the battlefield. However, to minimize unnecessary casualties to civilians, it is important to introduce additional checks and control from the control center before lethal force is authorized. Thus, the communication between drones and the control center becomes critical. In this thesis, we study the problem of communication between a military drone swarm and the control center when drones are flying over unfriendly territory where drones can be shot down by enemies. We present Hector, an algorithm based on machine learning, to minimize the overall casualty of drones by scheduling data transmission. Extensive simulations show that Hector significantly reduces casualties than traditional algorithms.

We look at robotic systems made of separate discrete components that, by self-assembling, can organize into physical structures of growing size. We review 22 such systems, exhibiting components ranging from passive mechanical parts to mobile robots. We present a taxonomy of the systems, and discuss their design and function. We then focus on a particular system, the swarm-bot. In swarm-bot, the components that assemble are self-propelled modules that are fully autonomous in power, perception, computation, and action. We examine the additional capabilities and functions self-assembly can offer an autonomous group of modules for the accomplishment of a concrete task: the transport of an object. The design of controllers is accomplished in simulation using techniques from biologically-inspired computing. We show that self-assembly can offer adaptive value to groups that compete in an artificial evolution based on their fitness in task performance. Moreover, we investigate mechanisms that facilitate the design of self-assembling systems. The controllers are transferred to the physical swarm-bot system, and the capabilities of self-assembly and object transport are extensively evaluated in a range of different environments. Additionally, the controller for self-assembly is transferred and evaluated on a different robotic system, a super-mechano colony. Given the breadth and quality of the results obtained, we can say that the swarm-bot qualifies as the current state of the art in self-assembling robots. Our work supplies some initial evidence (in form of simulations and experiments with the swarm-bot) that self-assembly can offer robotic systems additional capabilities and functions useful for the accomplishment of concrete tasks.

Groups of starlings can form impressive shapes as they travel northward together in the springtime. This is among a group of natural phenomena based on swarm behaviour. The research field of artificial intelligence in computer science, particularly the areas of robotics and image processing, has in recent decades given increasing attention to the underlying structures. The behaviour of these intelligent swarms has opened new approaches for face detection as well. G. Beni and J. Wang coined the term “swarm intelligence” to describe this type of group behaviour. In this context, intelligence describes the ability to solve complex problems. The objective of this project is to automatically find exactly one face on a photo or video material by means of swarm intelligence. The process developed for this purpose consists of a combination of various known structures, which are then adapted to the task of face detection. To illustrate the result, a 3D hat shape is placed on top of the face using an example application program.

Fire ants link their legs and jaws together to form functional structures called self- assemblages. Examples include floating rafts, towers, bridges, and bivouacs. We investigate these self-assemblages of fire ants. Our studies are motivated in part by the vision of providing guidance for programmable robot swarms. The goal for such systems is to develop a simple programmable element from which complex patterns or behaviors emerge on the collective level. Intelligence is decentralized, as is the case with social insects such as fire ants. In this combined experimental and theoretical study, we investigate the construction of two fire ant self-assemblages that are critical to the colony’s survival: the raft and the tower. Using time-lapse photography, we record the construction processes of rafts and towers in the laboratory. We identify and characterize individual ant behaviors that we consistently observe during assembly, and incorporate these behaviors into mathematical models of the assembly process. Our models accurately predict both the assemblages’ shapes and growth patterns, thus providing evidence that we have identified and analyzed the key mechanisms for these fire ant self-assemblages. We also develop novel techniques using scanning electron microscopy and micro-computed tomography scans to visualize and quantify the internal structure and packing properties of live linked fire ants. We compare our findings to packings of dead ants and similarly shaped granular material packings to understand how active arranging affects ant spacing and orientation. We find that ants use their legs to increase neighbor spacing and hence reduce their packing density by one-third compared to packings of dead ants. Also, we find that live ants do not align themselves in parallel with nearest neighbors as much as dead ants passively do. Our main contribution is the development of parsimonious mathematical models of how the behaviors of individuals result in the collective construction of fire ant assemblages. The models posit only simple observed behaviors based on local information, yet their mathe- matical analysis yields accurate predictions of assemblage shapes and construction rates for a wide range of ant colony sizes.

Submitted by Irene Nascimento (irene.kessia@ufpe.br) on 2016-08-24T17:35:05Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertacao-Natalia_Flora_de_Lima.pdf: 2000980 bytes, checksum: 107f0691d21b9d94e253d08f06a4fbdd (MD5) / Made available in DSpace on 2016-08-24T17:35:05Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertacao-Natalia_Flora_de_Lima.pdf: 2000980 bytes, checksum: 107f0691d21b9d94e253d08f06a4fbdd (MD5) Previous issue date: 2011-08-29 / Facepe / Este trabalho apresenta dois novos algoritmos, PSO-FPSO e FPSO-FPSO, para a otimização global de redes neurais MLP (do inglês Multi Layer Perceptron) do tipo feed-forward. O propósito destes algoritmos é otimizar de forma simultânea as arquiteturas e pesos sinápticos, objetivando melhorar a capacidade de generalização da rede neural artificial (RNA). O processo de otimização automática das arquiteturas e pesos de uma rede neural vem recebendo grande atenção na área de aprendizado supervisionado, principalmente em problemas de classificação de padrões. Além dos Algoritmos Genéticos, Busca Tabu, Evolução Diferencial, Recozimento simulado que comumente são empregados no treinamento de redes neurais podemos citar abordagens populacionais como a otimização por colônia de formigas, otimização por colônia de abelhas e otimização por enxame de partículas que vêm sendo largamente utilizadas nesta tarefa. A metodologia utilizada neste trabalho trata da aplicação de dois algoritmos do tipo PSO, sendo empregados na otimização das arquiteturas e na calibração dos pesos das conexões. Nesta abordagem os algoritmos são executados de forma alternada e por um número definido de vezes. Ainda no processo de ajuste dos pesos de uma rede neural MLP foram realizados experimentos com enxame de partículas heterogêneos, que nada mais é que a junção de dois ou mais PSOs de tipos diferentes. Para validar os experimentos com os enxames homogêneos foram utilizadas sete bases de dados para problemas de classificação de padrões, são elas: câncer, diabetes, coração, vidros, cavalos, soja e tireóide. Para os experimentos com enxames heterogêneos foram utilizadas três bases, a saber: câncer, diabetes e coração. O desempenho dos algoritmos foi medido pela média do erro percentual de classificação. Algoritmos da literatura são também considerados. Os resultados mostraram que os algoritmos investigados neste trabalho obtiveram melhor acurácia de classificação quando comparados com os algoritmos da literatura mencionados neste trabalho. / This research presents two new algorithms, PSO-FPSO e FPSO-FPSO, that can be used in feed-forward MLP (Multi Layer Perceptron) neural networks for global optimization. The purpose of these algorithms is to optimize architectures and synaptic weight, at same time, to improve the capacity of generalization from Artificial Neural Network (ANN). The automatic optimization process of neural network’s architectures and weights has received much attention in supervised learning, mainly in pattern classification problems. Besides the Genetic Algorithms, Tabu Search, Differential Evolution, Simulated Annealing that are commonly used in the training of neural networks we can mentioned population approaches such Ant Colony Optimization, Bee Colony Optimization and Particle Swarm Optimization that have been widely used this task. The methodology applied in this research reports the use of two PSO algorithms, used in architecture optimization and connection weight adjust. In this approach the algorithms are performed alternately and by predefined number of times. Still in the process of adjusting the weights of a MLP neural network experiments were performed with swarm of heterogeneous particles, which is nothing more than the joining of two or more different PSOs. To validate the experiments with homogeneous clusters were used seven databases for pattern classification problems, they are: cancer, diabetes, heart, glasses, horses, soy and thyroid. For the experiments with heterogeneous clusters were used three bases, namely cancer, diabetes and heart. The performance of the algorithms was measured by the average percentage of misclassification, literature algorithms are also considered. The results showed that the algorithms investigated in this research had better accuracy rating compared with some published algorithms.

A Wireless Sensor Network (WSN) is a group of spatially dispersed sensors that monitor the physical conditions of the environment and collect data at a central location. Sensor deployment is one of the main design aspects of WSNs as this a ffects network coverage. In general, WSN deployment methods fall into two categories: planned deployment and random deployment. This thesis considers planned sensor deployment of a Mobile Wireless Sensor Network (MWSN), which is defined as selectively deciding the locations of the mobile sensors under the given constraints to optimize the coverage of the network. Metaheuristic algorithms are powerful tools for the modeling and optimization of problems. The Krill Herd Algorithm (KHA) is a new nature-inspired metaheuristic algorithm which can be used to solve the sensor deployment problem. A Multi-Agent System (MAS) is a system that contains multiple interacting agents. These agents are autonomous entities that interact with their environment and direct their activity towards achieving speci c goals. Agents can also learn or use their knowledge to accomplish a mission. Multi-agent systems can solve problems that are very difficult or even impossible for monolithic systems to solve. In this work, a modification of KHA is proposed which incorporates MAS to obtain a Multi-Agent Krill Herd Algorithm (MA-KHA). To test the performance of the proposed method, five benchmark global optimization problems are used. Numerical results are presented which show that MA-KHA performs better than the KHA by finding better solutions. The proposed MA-KHA is also employed to solve the sensor deployment problem. Simulation results are presented which indicate that the agent-agent interactions in MA-KHA improves the WSN coverage in comparison with Particle Swarm Optimization (PSO), the Firefly Algorithm (FA), and the KHA. / Graduate

In my doctoral dissertation, I tackled two of the main open problems in swarm robotics: design and verification. I did so by using model checking.<p>Designing and developing individual-level behaviors to obtain a desired swarm-level goal is, in general, very difficult, as it is difficult to predict and thus design the non-linear interactions of tens or hundreds individual robots that result in the desired collective behavior. In my dissertation, I presented my novel contribution to the top-down design of robot swarms: property-driven design. Property-driven design is based on prescriptive modeling and model checking. Using property-driven design it is possible to design robot swarms in a systematic way, realizing systems that are "correct by design". I demonstrated property-driven design on two case-studies: aggregation and foraging.<p>Developing techniques to analyze and verify a robot swarm is also a necessary step in order to employ swarm robotics in real-world applications. In my dissertation, I explored the use of model checking to analyze and verify the properties of robot swarms. Model checking allows us to formally describe a set of desired properties of a system, in a more powerful and precise way compared to other mathematical approaches, and verify whether a given model of a system satisfies them. I explored two different approaches: the first based on Bio-PEPA and the second based on KLAIM. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished

In this dissertation, we propose and study methods for information transfer within a swarm of mobile robots that coordinately move, or flock, in a common direction. We define information transfer as the process whereby robots share directional information in order to coordinate their heading direction. We identify two paradigms of information transfer: explicit information transfer and implicit information transfer. <p><p>In explicit information transfer, directional information is transferred via communication. Explicit information transfer requires mobile robots equipped with a a communication device. We propose novel communication strategies for explicit information transfer, and we perform flocking experiments in different situations: with one or two desired directions of motion that can be static or change over time. We perform experiments in simulation and with real robots. Furthermore, we show that the same explicit information transfer strategies can also be applied to another collective behavior: collective transport with obstacle avoidance. <p><p>In implicit information transfer, directional information is transferred without communication. We show that a simple motion control method is sufficient to guarantee cohesive and aligned motion without resorting to communication or elaborate<p>sensing. We analyze the motion control method for its capability to achieve flocking with and without a desired direction of motion, both in simulation and using real robots. Furthermore, to better understand its underlying mechanism, we study this<p>method using tools of statistical physics, showing that the process can be explained in terms of non-linear elasticity and energy-cascading dynamics. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished

In this thesis, we investigate the problem of path formation and prey retrieval in a swarm of robots. We present two swarm intelligence control mechanisms used for distributed robot path formation. In the first, the robots form linear chains. We study three variants of robot chains, which vary in the degree of motion allowed<p>to the chain structure. The second mechanism is called vectorfield. In this case,<p>the robots form a pattern that globally indicates the direction towards a goal or<p>home location. Both algorithms were designed following the swarm robotics control<p>principles: simplicity of control, locality of sensing and communication, homogeneity<p>and distributedness.<p><p>We test each controller on a task that consists in forming a path between two<p>objects—the prey and the nest—and to retrieve the prey to the nest. The difficulty<p>of the task is given by four constraints. First, the prey requires concurrent, physical<p>handling by multiple robots to be moved. Second, each robot’s perceptual range<p>is small when compared to the distance between the nest and the prey; moreover,<p>perception is unreliable. Third, no robot has any explicit knowledge about the<p>environment beyond its perceptual range. Fourth, communication among robots is<p>unreliable and limited to a small set of simple signals that are locally broadcast.<p><p>In simulation experiments we test our controllers under a wide range of conditions,<p>changing the distance between nest and prey, varying the number of robots<p>used, and introducing different obstacle configurations in the environment. Furthermore,<p>we tested the controllers for robustness by adding noise to the different sensors,<p>and for fault tolerance by completely removing a sensor or actuator. We validate the<p>chain controller in experiments with up to twelve physical robots. We believe that<p>these experiments are among the most sophisticated examples of self-organisation<p>in robotics to date. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished

The intention of the dissertation is the applied research of the collective ( group ) ( swarm ) intelligence . To demonstrate the applicability of the collective intelligence, the Particle Swarm Optimization ( PSO ) algorithm has been studied in which the problem of the collective intelligence is transferred to mathematical optimization in which the particle swarm searches for a global optimum within the defined problem space, and the searching is controlled according to the pre-defined objective function which represents the solved problem. A new search strategy has been designed and experimentally tested in which the particles continuously adjust their behaviour according to the characteristics of the problem space, and it has been experimentally discovered how the impact of the objective function representing a solved problem manifests itself in the behaviour of the particles. The results of the experiments with the proposed search strategy have been compared to the results of the experiments with the reference version of the PSO algorithm. Experiments have shown that the classical reference solution, where the only condition is a stable trajectory along which the particle moves in the problem space, and where the influence of a control objective function is ultimately eliminated, may fail, and that the dynamic stability of the trajectory of the particle itself is not an indicator of the searching ability nor the convergence of the algorithm to the true global solution of the solved problem. A search strategy solution has been proposed in which the PSO algorithm regulates its stability by continuous adjustment of the particles behaviour to the characteristics of the problem space. The proposed algorithm influenced the evolution of the searching of the problem space, so that the probability of the successful problem solution increased.