Tomás Vidal Arredondo

Introductory Survey to Open-Source Mobile Robot Simulation Software

Mobile robot simulation is a valuable tool for education, research and design purposes. The last decade has seen a considerable increase in the development of new software tools for mobile robot simulation, all of which have reached different levels of maturity. This paper presents a survey of the existing tools and discusses their strengths and drawbacks in terms of simplicity, flexibility, fields of application, among other desirable features. This discussion should be valuable to students, teachers, engineers and researchers alike who are seeking adequate tools for simulating autonomous mobile robots. Introductory examples showing the usage of three of the most mature and widely used open-source simulators (Carmen, Gazebo and Open Dynamics Engine) are also included.

Feed-forward Artificial Neural Network based inference system applied in bioinformatics data-mining

This paper describes a neural network based inference system developed as part of a bioinformatic application in order to help implement a systematic search scheme for the identification of genes which encode enzymes of metabolic pathways. The inference system uses BLAST sequence alignment values as inputs and generates a classification of the best candidates for inclusion in a metabolic pathway map. The system considers a workflow that allows the user to provide feedback with their final classification decisions. These are stored in conjunction with analyzed sequences for re-training and constant inference system improvement.

Survey and comparative study of free simulation software for mobile robots

In robotics, simulation has become an essential tool for research, education, and design purposes. Various software tools for mobile robot simulation have been developed and have reached different levels of maturity in recent years. This paper presents a general survey of mobile robot simulation tools and discusses qualitative and quantitative aspects of selection of four major simulators publicly available at no cost: Carmen, Player-Stage-Gazebo, Open Dynamics Engine, and Microsoft Robotics Developer Studio. The comparison of the simulators is aimed at establishing the range of applications for which these are best suited as well as their accuracy for certain simulation tasks. The simulators chosen for detailed comparison were selected considering their level of maturity, modularity, and popularity among engineers and researchers. The qualitative comparison included a discussion of relevant features. The quantitative analysis entailed the development of a detailed dynamical model of a mobile robot on a road with varying slope. This model was used as benchmark to compare the accuracy of each simulator. The validity of the simulated results was also contrasted against measurements obtained from experiments with a real robot. This research and analysis should be very valuable to educators, engineers, and researchers who are always seeking adequate tools for simulating autonomous mobile robots.

Meta-learning based optimization of metabolic pathway data-mining inference system

This paper describes a novel meta-learning (MTL) based methodology used to optimize a neural network based inference system. The inference system being optimized is part of a bioinformatic application built to implement a systematic search scheme for the identification of genes which encode enzymes of metabolic pathways. Different MTL implementations are contrasted with manually optimized inference systems. The MTL based approach was found to be flexible and able to produce better results than manual optimization.

Meta-learning framework applied in bioinformatics inference system design

This paper describes a meta-learner inference system development framework which is applied and tested in the implementation of bioinformatic inference systems. These inference systems are used for the systematic classification of the best candidates for inclusion in bacterial metabolic pathway maps. This meta-learner-based approach utilises a workflow where the user provides feedback with final classification decisions which are stored in conjunction with analysed genetic sequences for periodic inference system training. The inference systems were trained and tested with three different data sets related to the bacterial degradation of aromatic compounds. The analysis of the meta-learner-based framework involved contrasting several different optimisation methods with various different parameters. The obtained inference systems were also contrasted with other standard classification methods with accurate prediction capabilities observed.

Serendipity in social networks

Serendipity is defined as fortunate discoveries made by chance. In this work we explore the idea that topological measures of a persons social network could be an indicator about how likely that person is to experience fortunate discoveries.

Fuzzy Motivations in Behavior Based Agents

In this chapter we describe a fuzzy logic based approach for providing biologically based motivations to be used by agents in evolutionary behavior learning. In this approach, fuzzy logic provides a fitness measure used in the generation of agents with complex behaviors which respond to user expectations of previously specified motivations. Our approach is implemented in behavior based navigation, route planning and action sequence based environment recognition tasks in a Khepera mobile robot simulator. Our fuzzy logic based motivation technique is shown as a simple and powerful method for agents to acquire a diverse set of fit behaviors as well as providing an intuitive user interface framework.

Learning performance in evolutionary behavior based mobile robot navigation

In this paper we utilize information theory to study the impact in learning performance of various motivation and environmental configurations. This study is done within the context of an evolutionary fuzzy motivation based approach used for acquiring behaviors in mobile robot exploration of complex environments. Our robot makes use of a neural network to evaluate measurements from its sensors in order to establish its next behavior. Adaptive learning, fuzzy based fitness and Action-based Environment Modeling (AEM) are integrated and applied toward training the robot. Using information theory we determine the conditions that lead the robot toward highly fit behaviors. The research performed also shows that information theory is a useful tool in analyzing robotic training methods.

Bio-inspired architecture for a reactive-deliberative robot controller

An incremental, bio-inspired control architecture for robots is proposed. The architecture is composed of two layers where each layer is intended to react in an analogous way to how our brain reacts to external and internal stimuli, using instinctive or deliberative reactions. The first layer, the Drive/Emotional Controller (DE Controller), uses Action Programs which are instinctual, low-level responses to a certain external stimuli. The second layer, the Deliberative/Reflective Controller (DR Controller), can only react to changes in the internal state of the DE controller. Memory, self-reflection, feelings and imagination are some of the concepts that could be considered by the DR Controller architecture, which is intended to generate high-level conscious reactions. These controllers integrate emotions into standard behavior-based architectures. We tested the proposed architecture in a task of free exploration using MODI, a two-wheeled mobile robot. Two DR controllers are tested in three different scenarios validating the proposed architecture in robotic applications.

Fuzzy Motivations in a Multiple Agent Behaviour-Based Architecture

In this article we introduce a blackboard-based multiple agent system framework that considers biologically-based motivations as a means to develop a user friendly interface. The framework includes a population-based heuristic as well as a fuzzy logic-based inference system used toward scoring system behaviours. The heuristic provides an optimization environment and the fuzzy scoring mechanism is used to give a fitness score to possible system outputs (i.e. solutions). This framework results in the generation of complex behaviours which respond to previously specified motivations. Our multiple agent blackboard and motivation-based framework is validated in a low cost mobile robot specifically built for this task. The robot was used in several navigation experiments and the motivation profile that was considered included “curiosity”, “homing”, “energy” and “missions”. Our results show that this motivation-based approach permits a low cost multiple agent-based autonomous mobile robot to acquire a diverse set of fit behaviours that respond well to user and performance expectations. These results also validate our multiple agent framework as an incremental, flexible and practical method for the development of robust multiple agent systems.