Fidel Aznar

L-system-driven self-assembly for swarm robotics

In this paper, an assembly swarm algorithm, that will generate microscopic rules from a macroscopic description of complex structures, will be presented. The global structure will be described in a formal way using L-systems (Lindenmayer systems). The proposed algorithm is mainly parallel and exhibit parsimony at microscopic level, being robust and adaptable. In addition, a comparation between a swarm with centralized control and our distributed swarm algorithm will be provided, comparing the time need by the swarm to be assembled and the number of messages exchanged between agents.

Agents for Swarm Robotics: Architecture and Implementation

Swarm robotics is a type of robotic systems based on many simple robots interactions. Such systems enjoy many benefits such as high tolerance and the possibility of increasing the number of robots in a transparent way to the programmer; but they also have many difficulties when applied to complex problems. In this paper, we will present a hybrid architecture for swarm robotics based on a multi-agent system. The main contribution of this architecture is to make possible the use of cognitive agents to lead a robotic swarm of simple agents without losing the advantages of swarms. Moreover, the implementation of this architecture within Real Swarm platform and the discussion of how to apply this architecture in real systems will be presented.

Using Gaussian Processes in Bayesian Robot Programming

In this paper, we present an adaptation of Gaussian Processes for learning a joint probabilistic distribution using Bayesian Programming. More specifically, a robot navigation problem will be showed as a case of study. In addition, Gaussian Processes will be compared with one of the most popular techniques for machine learning: Neural Networks. Finally, we will discuss about the accuracy of these methods and will conclude proposing some future lines for this research.

Obtaining a Bayesian map for data fusion and failure detection under uncertainty

This paper presents a generic Bayesian map and shows how it is used for the development of a task done by an agent arranged in an environment with uncertainty. This agent interacts with the world and is able to detect, using only readings from its sensors, any failure of its sensorial system. It can even continue to function properly while discarding readings obtained by the erroneous sensor/s. A formal model based on Bayesian Maps is proposed. The Bayesian Maps brings up a formalism where implicitly, using probabilities, we work with uncertainly. Some experimental data is provided to validate the correctness of this approach.

On Cooperative Swarm Foraging for Simple, Non Explicitly Connected, Agents

Nowadays there are several applications that use swarm robotics for solving research tasks and resource exploitation. Most of these applications are based on complex agents that require explicit communication between them. These systems are difficult to introduce in certain environments because of these features, where agents can not always communicate between them and where it would be necessary a large swarm. This paper presents a swarm system for a collective resource exploitation. The main features of the agents of this system are their simplicity and they do not communicate with each other in explicit way. A microscopic model that shows the individual performance of agents has been proposed, and a macroscopic model that describes the overall swarm system has been provided. Several tests that show the convergence of the swarm towards the best resource in an unknown environment have been analyzed.

Robot security and failure detection using bayesian fusion

This paper shows a Bayesian framework for fuse information. Using this framework we present a robotic system, based on two processing units. The system is used for the development of a task, done by an autonomous agent, arranged in an environment with uncertainty. This agent interacts with the world and is able to detect, only using its sensor readings, any failure of its sensorial system. Even it can continue working properly while discarding the readings obtained by the erroneous sensor/s. A security unit is also provided to make the system even more robust. The Bayesian Units brings up a formalism where implicitly, using probabilities, we work with uncertainly. Some experimental data are provided to validate the correctness of this approach.

A cognitive model for autonomous agents based on bayesian programming

This paper presents a cognitive model for an autonomous agent based on emotional psychology and Bayesian programming. A robot with emotional responses allows us to plan behaviour in a different way than present robotic architectures and provides us with a method of generating a new interface for human/robot interaction. The use of emotional modules means that the emotional state of the robot can be obtained directly and, therefore, it is relatively simple to obtain a virtual face that represents these emotions. An autonomous agent could have a model of the environment to be able to interact with the real universe where it is working. It is necessary to consider that any model of a real phenomenon will be incomplete due to the existence of uncertain, unknown variables that influence the phenomenon. Two example arquitectures are proposed here. Using these architectures some experimental data, to verify the correctness of this approach, is provided.

Visual Navigation for UAV with Map References Using ConvNets

In this paper, a visual system for helping unmanned aerial vehicles navigation, designed with a convolutional neural network, is presented. This network is trained to match on-board captured images with several previously obtained global maps, generating actions given a known global control policy. This system can be used directly for navigation or filtered, combining it with other aircraft systems. Our model will be compared with a classical map registration application, using a Scale-Invariant Feature Transform (SIFT) key point extractor. The system will be trained and evaluated with real aerial images. The results obtained show the viability of the proposed system and demonstrate its performance.

Learning Probabilistic Features for Robotic Navigation Using Laser Sensors

SLAM is a popular task used by robots and autonomous vehicles to build a map of an unknown environment and, at the same time, to determine their location within the map. This paper describes a SLAM-based, probabilistic robotic system able to learn the essential features of different parts of its environment. Some previous SLAM implementations had computational complexities ranging from O(Nlog(N)) to O(N 2), where N is the number of map features. Unlike these methods, our approach reduces the computational complexity to O(N) by using a model to fuse the information from the sensors after applying the Bayesian paradigm. Once the training process is completed, the robot identifies and locates those areas that potentially match the sections that have been previously learned. After the training, the robot navigates and extracts a three-dimensional map of the environment using a single laser sensor. Thus, it perceives different sections of its world. In addition, in order to make our system able to be used in a low-cost robot, low-complexity algorithms that can be easily implemented on embedded processors or microcontrollers are used.

A macroscopic model for high intensity radiofrequency signal detection in swarm robotics systems

In recent years, there has been a growing interest in resource location in unknown environments for robotic systems, which are composed of multiple simple robots rather than one highly capable robot [M. Sempere, F. Aznar, M. Pujol, and R. Rizo, On cooperative swarm foraging for simple, non explicitly connected, agents, 2010]. This tradeoff reduces the design and hardware complexity of the robots and removes single point failures, but adds complexity in algorithm design. The challenge is to programme a swarm of simple robots, with minimal intercommunication and individual capability, to perform a useful task as a group. This paper is focused on finding the highest intensity area of a radiofrequency (RF) signal in urban environments. These signals are usually more intense near the city centre and its proximity, since in these zones the risk of signal saturation is high. RF radiation (RFR) is boosted or blocked mainly depending on orography or building structures. RF providers need to supply enough coverage, setting up different antennas to be able to provide a minimum quality of service. We will define a micro/macroscopic mathematical model to efficiently study a swarm robotic system, predict their long-term behaviour and gain insight into the system design. The macroscopic model will be obtained from Rate Equations, describing the dynamics of the swarm collective behaviour. In our experimental section, the Campus of the University of Alicante will be used to simulate our model. Three RFR antennas will be taken into account, one inside our Campus and the other two in its perimeter. Several tests, that show the convergence of the swarm towards the RFR, will be presented. In addition, the obtained RFR maps and the macroscopic behaviour of the swarm will be discussed.