Ivan Villaverde

Two lattice computing approaches for the unsupervised segmentation of hyperspectral images

Endmembers for the spectral unmixing analysis of hyperspectral images are sets of affinely independent vectors, which define a convex polytope covering the data points that represent the pixel image spectra. Strong lattice independence (SLI) is a property defined in the context of lattice associative memories convergence analysis. Recent results show that SLI implies affine independence, confirming the value of lattice associative memories for the study of endmember induction algorithms. In fact, SLI vector sets can be easily deduced from the vectors composing the lattice auto-associative memories (LAM). However, the number of candidate endmembers found by this algorithm is very large, so that some selection algorithm is needed to obtain the full benefits of the approach. In this paper we explore the unsupervised segmentation of hyperspectral images based on the abundance images computed, first, by an endmember selection algorithm and, second, by a previously proposed heuristically defined algorithm. We find their results comparable on a qualitative basis.

Linked multi-component mobile robots: Modeling, simulation and control

The Linked Multi-Component Robotic Systems (L-MCRS) consists of a group of mobile robots carrying a passive uni-dimensional object (a hose or a wire). It is a recently identified unexplored and unexploited category of multi-robot systems. In this paper we report the first effort on the modeling, control and visual servoing of L-MCRS. Modeling has been tackled from geometrical and dynamical points of view. The passive element is modeled by splines, and the dynamical modeling is achieved by the appropriate extension of Geometrically Exact Dynamic Splines (GEDS). The systems modeling allows realistic simulation, which can be used as a test bed for the evaluation of control strategies. In this paper we evaluate two such control strategies: a baseline global controller, and a fuzzy local controller based on the observation of the hose segment between two robots. Finally, we have performed physical experiments on a team of robots carrying a wire under a visual servoing scheme that provides the perceptual information about the hose for the fuzzy local controller. Visual servoing robust image segmentation is grounded in the Dichromatic Reflection Model (DRM).

The Kosko Subsethood Fuzzy Associative Memory (KS-FAM): Mathematical Background and Applications in Computer Vision

Many well-known fuzzy associative memory (FAM) models can be viewed as (fuzzy) morphological neural networks (MNNs) because they perform an operation of (fuzzy) mathematical morphology at every node, possibly followed by the application of an activation function. The vast majority of these FAMs represent distributive models given by single-layer matrix memories. Although the Kosko subsethood FAM (KS-FAM) can also be classified as a fuzzy morphological associative memory (FMAM), the KS-FAM constitutes a two-layer non-distributive model.In this paper, we prove several theorems concerning the conditions of perfect recall, the absolute storage capacity, and the output patterns produced by the KS-FAM. In addition, we propose a normalization strategy for the training and recall phases of the KS-FAM. We employ this strategy to compare the error correction capabilities of the KS-FAM and other fuzzy and gray-scale associative memories in terms of some experimental results concerning gray-scale image reconstruction. Finally, we apply the KS-FAM to the task of vision-based self-localization in robotics.

Neuro-evolutionary mobile robot egomotion estimation with a 3D ToF camera

An innovative neuro-evolutionary approach for mobile robot egomotion estimation with a 3D ToF camera is proposed. The system is composed of two main modules following a preprocessing step. The first module is a Neural Gas network that computes a Vector Quantization of the preprocessed camera 3D point cloud. The second module is an Evolution Strategy that estimates the robot motion parameters by performing a registration process, searching on the space of linear transformations, restricted to the translation and rotation, between the codebooks obtained for successive camera readings. The fitness function is the matching error between the predicted and the observed codebook corresponding to the next camera readings. In this paper, we report results of an implementation of this system tested on data from a real mobile robot, and provide several comparisons between our and other well-known registration algorithms.

Convex Coordinates From Lattice Independent Sets for Visual Pattern Recognition

One of the key processes in nowadays intelligent systems is feature extraction. It pervades applications from computer vision to bioinformatics and data mining. The purpose of this chapter is to introduce a new feature extraction process based on the detection of extremal points on the cloud of points that represent the high dimensional data sample. These extremal points are assumed to define an approximation to the convex hull covering the data sample points. The features extracted are the coordinates of the data points relative to the extremal points, the convex coordinates. We have experimented this approach in several applications that will be summarized in the chapter.

Morphological Neural Networks for Localization and Mapping

Morphological associative memories (MAM) have been proposed for image denoising and pattern recognition. We have shown that they can be applied to other domains, like image retrieval and hyperspectral image unsupervised segmentation. In both cases the key idea is that morphological auto associative memories (MAAM) selective sensitivity to erosive and dilative noise can be applied to detect the morphological independence between patterns. The convex coordinates obtained by linear unmixing based on the sets of morphological independent patterns define a feature extraction process. These features may be useful either for pattern classification. We present some results on the task of visual landmark recognition for a mobile robot self-localization task

Lattice independent component analysis for mobile robot localization

This paper introduces an approach to appearance based mobile robot localization using Lattice Independent Component Analysis (LICA) The Endmember Induction Heuristic Algorithm (EIHA) is used to select a set of Strong Lattice Independent (SLI) vectors, which can be assumed to be Affine Independent, and therefore candidates to be the endmembers of the data Selected endmembers are used to compute the linear unmixing of the robots acquired images The resulting mixing coefficients are used as feature vectors for view recognition through classification We show on a sample path experiment that our approach can recognise the localization of the robot and we compare the results with the Independent Component Analysis (ICA).

Lattice independent component analysis for appearance-based mobile robot localization

This paper introduces an approach to appearance-based mobile robot localization using a new approach to dimensional reduction based on the notion of Lattice Independence called Lattice Independent Component Analysis (LICA). Any algorithm that can select a set of Strong Lattice Independent (SLI) vectors from the data can be applied inside LICA, this paper applies a specific Endmember Induction Algorithm (EIA) developed by our research group. The fact that SLI vectors are Affine Independent allows the coupling of non-linear Lattice Associative Memories (LAM) and linear unmixing for data exploration and dimensionality reduction. To perform an appearance-based mobile robot visual localization, images from the on-board camera robot are transformed into low dimension feature vector representations for classification. For validation, we compare LICA against several Independent Component Analysis (ICA) approaches over a collection of recorded image sequences taken from the robot following some predefined paths. Results show that LICA improves most of the ICA approaches, and it is only slightly improved by the Molgedey and Schouster ICA in some data instances.

An Improved Evolutionary Approach for Egomotion Estimation with a 3D TOF Camera

We propose an evolutionary approach for egomotion estimation with a 3D TOF camera. It is composed of two main modules plus a preprocessing step. The first module computes the Neural Gas (NG) approximation of the preprocessed camera 3D data. The second module is an Evolution Strategy which performs the task of estimating the motion parameters by searching on the space of linear transformations restricted to the translation and rotation, applied on the codevector sets obtained by the NG for successive camera readings. The fitness function is the matching error between the transformed last set of codevectors and the codevector set corresponding to the next camera readings. In this paper, we report new modifications and improvements of this system and provide several comparisons between our and other well known registration algorithms.

A Hybrid Intelligent System for Robot Ego Motion Estimation with a 3D Camera

A Hybrid Intelligent System (HIS) for self-localization working on the readings of innovative 3D cameras is presented in this paper. The system includes a preprocessing step for cleaning the 3D camera readings. The HIS consist of two main modules. First the Self-Organizing Map (SOM) is used to provide models of the preprocessed 3D readings of the camera. The 2D grid of the SOM units is assumed as a surface modeling the 3D data obtained from each snapshot of the 3D camera. The second module is an Evolution Strategy, which is used to perform the estimation of the motion of the robot between frames. The fitness function of the Evolution Strategy (ES) is given by the distance computed as the matching of the SOM unit grids.