Building Grounded Abstractions for Artificial Intelligence Programming

Most Artificial Intelligence (AI) work can be characterized as either ``high-level'' (e.g., logical, symbolic) or ``low-level'' (e.g., connectionist networks, behavior-based robotics). Each approach suffers from particular drawbacks. High-level AI uses abstractions that often have no relation to the way real, biological brains work. Low-level AI, on the other hand, tends to lack the powerful abstractions that are needed to express complex structures and relationships. I have tried to combine the best features of both approaches, by building a set of programming abstractions defined in terms of simple, biologically plausible components. At the ``ground level'', I define a primitive, perceptron-like computational unit. I then show how more abstract computational units may be implemented in terms of the primitive units, and show the utility of the abstract units in sample networks. The new units make it possible to build networks using concepts such as long-term memories, short-term memories, and frames. As a demonstration of these abstractions, I have implemented a simulator for ``creatures'' controlled by a network of abstract units. The creatures exist in a simple 2D world, and exhibit behaviors such as catching mobile prey and sorting colored blocks into matching boxes. This program demonstrates that it is possible to build systems that can interact effectively with a dynamic physical environment, yet use symbolic representations to control aspects of their behavior.

Active Learning with Statistical Models

For many types of learners one can compute the statistically 'optimal' way to select data. We review how these techniques have been used with feedforward neural networks. We then show how the same principles may be used to select data for two alternative, statistically-based learning architectures: mixtures of Gaussians and locally weighted regression. While the techniques for neural networks are expensive and approximate, the techniques for mixtures of Gaussians and locally weighted regression are both efficient and accurate.

Control de posició del robot wifibot

Aquest projecte pretén presentar de forma clara i detallada l’estructura i el funcionament del robot així com dels components que el conformen. Aquesta informació és de vital importància a l’hora de desenvolupar aplicacions per al robot. Un cop descrites les característiques del robot s’analitzaran les eines necessàries i/o disponibles per poder desenvolupar programari per cada nivell de la forma més senzilla i eficient possible. Posteriorment s’analitzaran els diferents nivells de programació i se’n contrastaran els avantatges i els inconvenients de cada un. Aquest anàlisi es començarà fent pel nivell més alt i anirà baixant amb la intenció de no entrar en nivells més baixos del necessari. Baixar un nivell en la programació suposa haver de crear aplicacions sempre compatibles amb els nivells superiors de forma que com més es baixa més augmenta la complexitat. A partir d’aquest anàlisi s’ha arribat a la conclusió que per tal d’aprofitar totes les prestacions del robot és precís arribar a programar en el nivell més baix del robot. Finalment l’objectiu és obtenir una sèrie de programes per cada nivell que permetin controlar el robot i fer-lo seguir senzilles trajectòries

Integració i estudi d’un conjunt de propulsors per un robot submarí

L’objectiu d’aquest projecte/treball fi de carrera es estudiar els propulsors i el seu protocol de comunicació proporcionant informació útil a l’hora de dissenyar i construir el robot subaquàtic que implementi els propulsors

Iniciación del alumno al campo de la investigación en una asignatura multidisciplinar orientada a la robótica

La asignatura de libre elección en la que se realizó la experiencia que se indica pretende ser un trampolín de iniciación para aquellos alumnos de la Universidad Europea de Madrid que tengan inquietud por descubrir el mundo de la robótica. Con una clase de alumnos procedentes de muchas titulaciones distintas, Licenciado en Odontología, Ingeniero de Caminos, Canales y Puertos, Ingeniero Industrial, Ingeniero Informático, Ingeniero en Telecomunicaciones, Técnicos en Obras Públicas y alumnos internacionales, el reto de hacer de la asignatura algo interesante para ellos implicaba saber adaptarse a distintos niveles tanto disciplinar (varias carreras) como académico (los alumnos eran tanto de los primeros cursos como de los últimos). Basándose en el uso del portafolio y el aprendizaje basado en problemas se fueron inculcando los conocimientos básicos necesarios para desarrollar lo que sería el final de la asignatura. Este objetivo final es el que hizo que los alumnos vieran de cerca la labor de un investigador y un grupo de trabajo multidisciplinar. El reto consistió en que debían hacer una solicitud 'ficticia' de un proyecto PROFIT. Los PROFIT constituyen programas de ayuda y fomento a la investigación técnica convocados por el Ministerio de Industria, Turismo y Comercio. Las plantillas son accesibles desde Internet y de esta forma los alumnos pudieron realizar una memoria clara y precisa de sus proyectos. Además, como elemento final de evaluación se invitó a dos profesores expertos en robótica de otra universidad al día de la presentación en el que los alumnos entregaban la memoria y defendían sus trabajos. Tres profesores en total, dos de otra universidad y el profesor de la asignatura asistieron a su defensa y pusieron de manera independiente los trabajos en orden según sus preferencias, al ser 5 grupos la nota debía ponerse entre 10, 9, 8, 7 y 6. La media de la decisión de los tres profesores configuró la nota final

Estudio de la extracción química del fe en madera arqueológica subacuatica tratada previamente con peg 4000

Debido a la gran cantidad de muestras arqueológicas impregnadas con PEG que se encuentran contaminadas por compuestos insolubles de hierro, se plantea la posible extracción y formación de complejos Fe-L (L=PBTC) y sus efectos en (i) la estructura de la matriz orgánica, (ii) la estructura y propiedades físicas del PEG y (iii) el comportamiento de la muestra en la etapa posterior de almacenamiento. El proyecto analiza la formación de compuestos químicos y posibles modificaciones estructurales en el proceso de extracción del hierro. Consiste en un estudio sistemático de un sistema químico y su influencia en los procesos de precipitación de Fe3+ en medio acuoso. El proyecto se fundamenta en: (1) desarrollar un proceso experimental de optimización para la extracción de las sales contaminantes y (2) encontrar las técnicas analíticas óptimas que permitan apreciar modificaciones estructurales de los diferentes sistemas. Se determina la cantidad de hierro extraído mediante A.A. Las interacciones entre PBTC y PEG se analizan por IR. Las modificaciones de determinadas propiedades físicas se determinan por DSC y las estructurales mediante SEM. En las condiciones termodinámicas óptimas se obtiene una extracción superficial del hierro (30-35%). La disolución del PEG origina modificaciones de la masa y el volumen de la muestra

Immersió!

Vint-i-un estudiants de l'IES Castell d'Estela d'Amer han participat durant tres dies en una experiència de recerca amb els investigadors del centre ViCOROB de la UdG

An hybrid methodology for RL-based behavior coordination in a target following mission with an AUV

Proposes a behavior-based scheme for high-level control of autonomous underwater vehicles (AUVs). Two main characteristics can be highlighted in the control scheme. Behavior coordination is done through a hybrid methodology, which takes in advantages of the robustness and modularity in competitive approaches, as well as optimized trajectories

Colour image processing

In the context of the round table the following topics related to image colour processing will be discussed: historical point of view. Studies of Aguilonius, Gerritsen, Newton and Maxwell. CIE standard (Commission International de lpsilaEclaraige). Colour models. RGB, HIS, etc. Colour segmentation based on HSI model. Industrial applications. Summary and discussion. At the end, video images showing the robustness of colour in front of B/W images will be presented

Hybrid coordination of reinforcement learning-based behaviors for AUV control

This paper proposes a hybrid coordination method for behavior-based control architectures. The hybrid method takes advantages of the robustness and modularity in competitive approaches as well as optimized trajectories in cooperative ones. This paper shows the feasibility of applying this hybrid method with a 3D-navigation to an autonomous underwater vehicle (AUV). The behaviors are learnt online by means of reinforcement learning. A continuous Q-learning implemented with a feed-forward neural network is employed. Realistic simulations were carried out. The results obtained show the good performance of the hybrid method on behavior coordination as well as the convergence of the behaviors

Efficient learning of reactive robot behaviors with a Neural-Q_learning approach

The purpose of this paper is to propose a Neural-Q_learning approach designed for online learning of simple and reactive robot behaviors. In this approach, the Q_function is generalized by a multi-layer neural network allowing the use of continuous states and actions. The algorithm uses a database of the most recent learning samples to accelerate and guarantee the convergence. Each Neural-Q_learning function represents an independent, reactive and adaptive behavior which maps sensorial states to robot control actions. A group of these behaviors constitutes a reactive control scheme designed to fulfill simple missions. The paper centers on the description of the Neural-Q_learning based behaviors showing their performance with an underwater robot in a target following task. Real experiments demonstrate the convergence and stability of the learning system, pointing out its suitability for online robot learning. Advantages and limitations are discussed

Semi-online neural-Q_leaming for real-time robot learning

Reinforcement learning (RL) is a very suitable technique for robot learning, as it can learn in unknown environments and in real-time computation. The main difficulties in adapting classic RL algorithms to robotic systems are the generalization problem and the correct observation of the Markovian state. This paper attempts to solve the generalization problem by proposing the semi-online neural-Q_learning algorithm (SONQL). The algorithm uses the classic Q_learning technique with two modifications. First, a neural network (NN) approximates the Q_function allowing the use of continuous states and actions. Second, a database of the most representative learning samples accelerates and stabilizes the convergence. The term semi-online is referred to the fact that the algorithm uses the current but also past learning samples. However, the algorithm is able to learn in real-time while the robot is interacting with the environment. The paper shows simulated results with the "mountain-car" benchmark and, also, real results with an underwater robot in a target following behavior

Towards Direct Policy Search Reinforcement Learning for Robot Control

This paper proposes a high-level reinforcement learning (RL) control system for solving the action selection problem of an autonomous robot. Although the dominant approach, when using RL, has been to apply value function based algorithms, the system here detailed is characterized by the use of direct policy search methods. Rather than approximating a value function, these methodologies approximate a policy using an independent function approximator with its own parameters, trying to maximize the future expected reward. The policy based algorithm presented in this paper is used for learning the internal state/action mapping of a behavior. In this preliminary work, we demonstrate its feasibility with simulated experiments using the underwater robot GARBI in a target reaching task

A system to evaluate the accuracy of a visual mosaicking methodology

When underwater vehicles navigate close to the ocean floor, computer vision techniques can be applied to obtain motion estimates. A complete system to create visual mosaics of the seabed is described in this paper. Unfortunately, the accuracy of the constructed mosaic is difficult to evaluate. The use of a laboratory setup to obtain an accurate error measurement is proposed. The system consists on a robot arm carrying a downward looking camera. A pattern formed by a white background and a matrix of black dots uniformly distributed along the surveyed scene is used to find the exact image registration parameters. When the robot executes a trajectory (simulating the motion of a submersible), an image sequence is acquired by the camera. The estimated motion computed from the encoders of the robot is refined by detecting, to subpixel accuracy, the black dots of the image sequence, and computing the 2D projective transform which relates two consecutive images. The pattern is then substituted by a poster of the sea floor and the trajectory is executed again, acquiring the image sequence used to test the accuracy of the mosaicking system

Towards a real-time vision-based navigation system for a small-class UUV

This paper deals with the problem of navigation for an unmanned underwater vehicle (UUV) through image mosaicking. It represents a first step towards a real-time vision-based navigation system for a small-class low-cost UUV. We propose a navigation system composed by: (i) an image mosaicking module which provides velocity estimates; and (ii) an extended Kalman filter based on the hydrodynamic equation of motion, previously identified for this particular UUV. The obtained system is able to estimate the position and velocity of the robot. Moreover, it is able to deal with visual occlusions that usually appear when the sea bottom does not have enough visual features to solve the correspondence problem in a certain area of the trajectory