19 research outputs found
Impressionist Hole Detection and Healing Using Swarms of Agents with Quantized Perception
No full textCoverage holes are a key problem in wireless sensor
networks. Methods that use relative localization techniques to
restore the service, or heal the holes, rely on accurate range
and bearing measurements. However, high-precision range and
bearing sensors are too heavy, expensive, and range-limited for
the agents tasked with healing. To overcome these limitations,
we propose a novel impressionist algorithm, inspired by a recent
swarm-based approach, that works with extremely quantized
range and bearing information and at low perception frequency,
to detect and heal the holes. In the proposed approach, a swarm
of agents navigate using quantized bearing information in a
potential field generated by nodes to reach the nearest hole.
The swarm adopts a greedy deployment behavior, preventing
concurrent placements in close-by locations. After deployment,
agents use their coarse perception to update the potential field,
guiding the rest of the swarm to unhealed area. Simulation
results demonstrate that our algorithm achieves similar or better
coverage compared to the state-of-the-art and to a benchmark
based on random walk. This is achieved using just three bearing
quantization levels and four times lower perception frequency.
Overall, our impressionist approach shows faster healing, albeit
at the expense of employing slightly more agents
Using Artificial Immune System to Prioritize Swarm Strategies for Environmental Monitoring
No full text
Q-learning-based distributed algorithm for hole recovery in Wireless Sensor Networks
No full textIn this work, the implementation of a distributed algorithm based on Q-learning for hole recovery in Wireless Sensor Networks is described. Initially, a detailed analysis of the state-of-the-art for Hole Detection and Recovery has been carried out, focusing on Reinforcement Learning approaches. Then, a simulation environment for multi-agent reinforcement learning hole recovery algorithms has been implemented in Matlab. Subsequently, a state-of-the-art Q-learning based algorithm has been developed in the simulation environment, adding also a decaying rate for the exploration rate. Finally, Differential Evolution has been used for Hyperparameters Optimization
Development in autopilot environment of a swarm robotics algorithm for hole detection in WSN
No full textThis thesis exploresinto the development of hole detection algorithms for swarm robotics, with a focus on advanced simulation environments. It explores the integration and comparative analysis of PX4 Autopilot and the JMAVSim-Gazebo combination as essential tools for evaluating and refining autonomous aerial systems. The research also introduces a novel "Localized Movement-Assisted Sensor Deployment Algorithm for Hole Detection and Healing" in the realm of wireless sensor networks. This algorithm leverages movement-assisted deployment to address the coverage hole problem and has shown superior performance in terms of accuracy, efficiency, and energy consumption. The practical implications of these domains span various real-world applications, emphasizing the importance of advanced simulation environments in technology development.
In this comprehensive investigation, i explored into the implementation of the sensor deployment algorithm, showcasing the transformative potential of autonomous drones. Real-case scenarios and simulations within the Gazebo Classic Simulator highlight the adaptability and precision of drone-based systems. The research contributes to the dynamic field of drone technology, emphasizing the role of efficient sensor deployment strategies in practical applications across industries
A Comparative Analysis of Artificial Potential Fields' Shapes for Lattice Formation in Swarm Robotics
No full textThe artificial potential fields method stands as a common methodology for guiding autonomous robots behavior. Despite its wide use, a critical gap exists in the literature regarding the impact of the potentials shape, especially in the context of swarm robotics. Five attractive-repulsive artificial potentials commonly used in literature, together with a newly introduced potential, have been experimentally evaluated for their ability to solve lattice formation in simulated robotic swarms under varying conditions of noise and levels of cognitive speed. Results highlight the crucial impact of potential shape in swarm tasks. Potentials that are not diverging and non-smooth around the equilibrium point are poorly effective even for small swarm sizes. Smooth but non-diverging potentials exhibit acceptable yet sub-optimal performance. Smooth and diverging potentials like Lennard-Jones consistently proved to be optimal across the vast majority of experimental conditions
Sviluppo di una piattaforma di uso industriale composta da software autopilota e simulatore open source per la gestione di missioni basate su droni
No full textn primo luogo, è stata fatta una descrizione generale del mondo PX4, delle principali
caratteristiche e al lettore vengono descritte le peculiarità di questo sistema e una breve
guida del protocollo di comunicazione da questo utilizzato.
Dopodichè è stata fatta una descrizione di una possibile API OFFBOARD utilizzabile
per l’interazione con dispositivi PX4 al fine di eseguire alcune missioni e del simulatore
da poter utilizzare per provare inizialmente su PC il comportamento del codice scritto.
Nell’ultima parte della tesi c’è l’effettiva implementazione della missione di ricerca
mostrando brevemente tutte le fasi necessarie per la realizzazione. Infine, i risultati
Hole detection and recovering algorithms using bio-inspired robotic swarms
No full textA Wireless Sensor Network (WSN) is a set of spatially distributed autonomous sensors with sensing, computation and wireless communication capabilities. These networks are used in many fields such as national security, surveillance, health care, biological detection and environmental monitoring. However, these sensors are characterized by limited wireless and computing capabilities and therefore they should be carefully deployed in order to cover the areas to be monitored without impairing network lifetime.
Unfortunately, these, as all the kinds of networks, are characterized by a normal and not isolated event: fault. Thus, in order to guarantee the network quality of service it is essential for the WSNs to be able to detect failure and perform akin to healing, recovering from events that might cause some of its part malfunction.
We focused on networks where the sensors have the ability to move in the space as the Flying ad-hoc Networks (FANETs) where the monitoring sensors are mounted on top of UAVs (Unmanned Air Vehicles) and propose a framework for hole detection and recovering technique from the malfunctioning inspired from how platelets work in recovering a wound
Superfície polinomial de resposta num ensaio de adubação com níveis não equidistantes
No full textA regressão polinomial é comumente aplicada em ensaios de adubação, mas quase sempre com níveis equidistantes de fertilizantes. Neste trabalho estudamos um ensaio fatorial de 3 x 3 x 3 com N, P e K, em níveis não equidistantes, que. eram os seguintes: N: zero, 30, 50 kg/ha, P2O5: zero, 45, 60 kg/ha, K2O. Parte da teoria, mostrada para o caso de níveis igualmente espaçados, foi adaptada para sua aplicação no caso estudado. A equação de regressão obtida foi: (Descrito na Tese). Os intervalos confiança obtidos de para os parâmetros foram relativamente grandes, o que concorda com grande parte dos trabalhos constantes da bibliografia. As estimativas das produções obtidas através da equação de regressão foram relativamente boas, com intervalos de confiança bem pequenos. A análise de variância mostrou efeito altamente significativo para a regressão. O coeficiente de determinação foi de 0,949, ou seja, 94,9% da variação foi explicada pela regressão. Os testes dos parâmetros foram altamente significativos em todos os casos, coro exceção para os parâmetros a22 e a33. Isto mostra, que para P e K a produção cresce de modo aproximadamente linear dentro dos limites do ensaio. A receita liquida, dada pela equação: Z= wŷ - t1 x1 - t2 x2 - t3 x1 - m foi também estudada, com os valores seguintes: w (preço do milho): 0,42 bolivares/kg, t1 (preço do N): 13,6 bolivares/kg, t2 (preço do P2O5): 7,0 bolivares/kg, t<sub1 (preço do K2O): 4,4 bolivares/kg. Esta função, para o ensaio estudado, não possui máximo, mas sim um ponto de sela. Entretanto, se considerarmos somente os valores contidos dentro dos intervalos usados no experimento, há um máximo absoluto na função da receita liquida, para x1= 4,94, x2= 6,00 x3= 5,00. Assim podemos recomendar no presente caso, as seguintes doses de nutrientes: 49,4 kg/ha de N, 60,0 kg/ha de P2O5, 50,0 kg/ha de K2O.Polynomial regression is commonly appllied in the and.lysis of fertilizer experiments, but almost always with equally spaced levels. The author studies a 3x3x3 N, P, K factorial trial with unequally spaced levels, which were the followingg: N: zero, 30, 50 kg/ha, P2O5: zero, 45, 60 kg/ha, K2O. First of all the theory, shown for the case of equally spaced levels, was modified in order to be applied to the case under study. The regres.sion equation obtained was: (See Thesis). The confidence intervals for the parameters were relatively large, which is in good agreement with many of the references. The harvest estimates obtained by the use of the regression equation were good, with rather short confidence intervals. The analysis of variance gave a highly significant F for regression. The coefficient of determination was 0.949, that is, 94.9% of the variation were explained by the regression equation. Tests for the parameters were highly significant in all cases, except for a22 and a33. This shows, of course, that for P and K the response is approxirnately linear within the limits of fertilization ín the experiment. The net income, give by equacion: Z= wŷ - t1 x1 - t2 x2 - t3 x1 - m, was studied also, with the following values: w (price of maize): 0,42 bolivares/kg, t1 (price of N): 13,6 bolivares/kg, t2 (price of P2O5): 7,0 bolivares/kg, t<sub1 (price of K2O): 4,4 bolivares/kg. The net income function has no ma:ximum, but a saddle point. However, if we consider only the values within the intervals used in the expe- riment, there is an absolute maximum income for x1= 4,94, x2= 6,00 x3= 5,00. So, we should recommend, in the present case, the following levels of nutrients: 49,4 kg/ha de N, 60,0 kg/ha de P2O5, 50,0 kg/ha de K2O
