Integrated Computer-Aided Engineering

Cognitive Developmental Robotics relies on lifelong open-ended learning processes, where mechanisms are needed to allow the robot to self-discover and self-select goals as well as to self-define its state space evaluation with regards to them. Thus, this paper addresses the problem of finding and using goals in continuous state spaces and automatically obtaining sub-goal hierarchies that allow autonomous development. In particular, the main purpose of this paper is to propose a new approach to the creation of utility models based on the concept of separable utility regions (SURs), which reduce the complexity of standard value function like utility models. These regions exhibit a correlation between the expected utility and the response of one sensor of the robot. Once they are discovered, the evaluation of the candidate states is only based on the changes of one sensor, which provides a strong independence from noise or dynamism in the utility models. A non-static variation of the classical collect-a-ball scenario and a robot gathering problem were used to test this approach in simulation and on real robots in order to identify goals and sub-goals in an autonomous way. The results confirm the good response of the method as a highly promising approach towards autonomous learning of continuous domains in cognitive robotics.

Pan-tilt-zoom (PTZ) cameras are well suited to motion detection and tracking objects due to their mobility. Motion detection approaches based on background difference have been the most used with fixed cameras because of the high quality of the achieved segmentation. However, time requirements and high costs prevent most of the algorithms proposed in literature from exploiting the background difference with PTZ cameras in real world applications, such as automatic surveillance. This paper presents a new algorithm to detect moving objects within an area covered by a PTZ camera while it is panning, tilting or zooming in or out. The low computational demands of the algorithm allow for its deployment to a Raspberry Pi microcontroller-based board, which enables the design and implementation of a low-cost monitoring system that is able to perform real-time image processing. First, our system works offline to estimate the parameters of the motion detection model, which are written on the Raspberry Pi memory. Second, motion detection is performed online by the microcontroller. Experimental results using different moving objects classifiers (FANN, KNN, and SVM) confirm the good performance of this approach in terms of different classification performance measures (accuracy, F-measure, AUC, and sample processing time).

Using Context-Awareness information in human-robot shared environments enhances the characterization of the interaction scenario, improving the experience of human collaboration with the robot. Aspects such as environmental location and human-robot dialog have traditionally been used to infer the current situation, which can be helpful in the decision making process of any robot. However, the acoustic signals available in every interaction scenario are commonly obviated, thus removing an important source of information. This paper presents the design, development and tests of a Context Awareness Component that labels users’ activities using localization information, dialog flow, and time of day, and adds an environment recognition component supported by acoustic signals that improves the inference system. To this end, this research proposes a feedforward neural network solution based on a multilayer perceptron approach. The paper also discusses the configuration of the neural network for optimizing the recognition of human activity in “at home” environments, using the four inputs previously mentioned. Finally, the validation of the approach proposed is done by comparing the results when the sound recognition system is used and when it is not.

It seems clear that general adoption of electric vehicles is coming in the near future. But this adoption will bring new challenges as, for example, that of recharging the batteries of a large fleet of electric vehicles under power and other technological constraints of the charging infrastructure. Among others, these will require solving challenging scheduling problems as well. In this paper, we study one of such problems derived from a charging station designed to be installed in community parks, which consists in scheduling a set of jobs on a single machine with varying capacity over time and exhibits high computational complexity. We propose the use of meta-heuristics as a means to solving the problem efficiently. Concretely, we propose a memetic algorithm, that combines a genetic algorithm with a local search method specifically designed for the problem. The contributions are analyzed theoretically, with formal proofs of their properties, and evaluated empirically. Experimental results show that the proposed memetic algorithm is very effective at solving the problem, while keeping running times reasonably low.

This paper tackles the job shop scheduling problem with fuzzy sets modelling uncertain durations and flexible due dates. The objective is to achieve high-service level by maximising due-date satisfaction, considering two different overall satisfaction measures as objective functions. We show how these functions model different attitudes in the framework of fuzzy multicriteria decision making and we define a measure of solution robustness based on an existing a-posteriori semantics of fuzzy schedules to further assess the quality of the obtained solutions. As solving method, we improve a memetic algorithm from the literature by incorporating a new heuristic mechanism to guide the search through plateaus of the fitness landscape. We assess the performance of the resulting algorithm with an extensive experimental study, including a parametric analysis, and a study of the algorithm’s components and synergy between them. We provide results on a set of existing and new benchmark instances for fuzzy job shop with flexible due dates that show the competitiveness of our method.

Robust identification and tracking of the pupil provides key information that can be used in several applications such as controlling gaze-based HMIs (human machine interfaces), designing new diagnostic tools for brain diseases, improving driver safety, detecting drowsiness, performing cognitive research, among others. We propose a deep convolutional neural network for eye-tracking based on atrous convolutions and spatial pyramids. DeepEye is able to handle real world problems such as varying illumination, blurring and reflections. The proposed network was trained and evaluated on 94,000 images taken from 24 data sets recorded in real world scenarios. DeepEye outperforms previous eye-tracking methods tested with these data sets. It improves the results of the current state of the art in a 26%, achieving an accuracy of more than 70% in almost every data set in terms of percentage of pupils detected with a distance error lower than 5 pixels. DeepEye can be downloaded at: https://github.com/Fjaviervera/DeepEye.

Human-Robot Interaction (HRI) is a growing area of interest in Artificial Intelligence that aims to make interaction with robots more natural. In this sense, numerous research studies on verbal and visual interactions with robots have appeared. The present paper will focus on non-verbal communication and, more specifically, gestures related to speech, which is an open question. With the aim of developing this part of Human-Robot Interaction or HRI, a new architecture is proposed for the assignment of gestures to speech based on the analysis of semantic similarities. In this way, gestures will be intelligently selected using Natural Language Processing (NLP) techniques. The conditions for gesture selection will be determined from an assessment of the effectiveness of different language models in a lexical substitution task applied to gesture annotation. On the basis of this analysis, the aim is to compare models based on expert knowledge and statistical models generated from lexical learning.

This paper provides a novel method to design a Rainfall Forecasting System based on Neural Networks and a long-term data registry. The System proposed is based on the observations of expert meteorologists after 40 years of research developing local forecast methodologies to predict rainfall events in the Meteorological Observatory of Valladolid, Spain. The Geostrophic Wind, a theoretical value resulting of the balance between the Pressure Gradient Force and the Coriolis Force, is the key parameter which feeds the aforementioned System. The paper focuses on the Geostrophic Wind calculation. We propose a novel method to estimate its direction and magnitude on a local scale, by processing a set of pressure and temperature observations surrounding a central location. This study continues previous research, providing complementary criteria for numerical weather prediction systems based on time-series forecasting and neural networks.