Advanced Topics on Computer Vision, Control and Robotics in Mechatronics

Medical ultrasound images are inherently contaminated by a multiplicative noise called speckle. The noise reduces the resolution and contrast, decreasing the capability of the visual evaluation of the image, and sometimes small speckles can mask ills in early stages. Therefore, denoising plays an important role in the diagnostic. Many investigations reported in the literature claim their performance. However, this is limited because the unclear indicators or sometimes the algorithms proposed are not suitable for implementations in hardware. In this chapter, the implementation of five methods, specifically designed to reduce multiplicative noise, in a digital signal processor is presented. The chapter includes performance evaluation of each method implemented in a fixed point, DM6437 digital signal processor (digital media processor) of Texas Instruments™. Results show that the performance of the Frost and Lee filters, with a local window of 5 × 5 pixels, is better to reduce high-variance speckle noise than the rest of the filters. For noise variance less than 0.1, the SRAD with 15 iterations has a higher performance. However, the Frost and SRAD filters take more time to yield a result.

Morphological neural networks, in particular, those with dendritic processing (MNNDPs), have shown to be a very promising tool for pattern classification. In this chapter, we present a survey of the most recent advances concerning MNNDPs. We provide the basics of each model and training algorithm; in some cases, we present simple examples to facilitate the understanding of the material. In all cases, we compare the described models with some of the state-of-the-art counterparts to demonstrate the advantages and disadvantages. In the end, we present a summary and a series of conclusions and trends for present and further research.

In this chapter, a mobile augmented reality prototype to support the process of manufacturing an all-terrain vehicle (ATV) is presented. The main goal is assisting the automotive industry in the manufacturing process regarding vehicle design and new model’s introduction; in addition, the activities of training and quality control can be supported. The prototype is composed of three main stages: (a) welding inspection, (b) measuring of critical dimensions inspection, and (c) mounting of virtual accessories in the chassis. A set of 3D models and 2D objects was used as virtual elements related to augmented reality. The prototype was tested regarding usability in a real industrial stage by measuring the scope of markers’ detection and by means of a survey. The results obtained demonstrated that the prototype is useful for the manufacturing of an ATV.

Pattern recognition is not a new field, but the challenges are coming on the data format. Today’s technological devices provide a huge amount of data with extensive detail evolving the classical pattern recognition approaches for dealing with them. Given the size of and quantity of descriptors data possess, traditional pattern recognition techniques have to draw on feature selection to handle problems like the excess of computer resources and dimensionality. Feature selection techniques are evolving, as well, for data related reasons. Chronologically linked data brings new challenges to the field. In the present chapter, we expose the gap in feature selection research to handle this type of data, as well as give suggestions of how to perform or pursue an approach to chronologically linked data feature selection.

In the last three decades, multi-frame and single-frame super-resolution and reconstruction techniques have been receiving increasing attention because of the large number of applications that many areas have found when increasing the resolution of their images. For example, in high-definition television, high-definition displays have reached a new level and resolution enhancement cannot be ignored; in some remote sensing applications, the pixel size is a limitation; and in medical imaging, the details are important for a more accurate diagnostic or acquiring high-resolution images while reducing the time of radiation to a patient. Some of the problems faced in this area, that in the future require dealing more effectively, are the inadequate representation of edges, inaccurate motion estimation between images, sub-pixel registration, and computational complexity among others. In this chapter, an overview of the most important methods classified into two taxonomies, multiple- and single-image super-resolution, is given. Moreover, two new techniques for single-image SR are proposed.

Cognitive robotics has focused its attention on the design and construction of artificial agents that are able to perform some cognitive task autonomously through the interaction of the agent with its environment. A central issue in these fields is the process of learning. In its attempt to imitate cognition in artificial agents, cognitive robotics has implemented models of cognitive processes proposed in areas such as biology, psychology, and neurosciences. A novel methodology for the control of autonomous artificial agents is the paradigm that has been called neuro-robotics or embedded neural cultures, which aims to embody cultures of biological neurons in artificial agents. The present work is framed in this paradigm. In this chapter, simulations of an autonomous learning process of an artificial agent controlled by artificial action potential neural networks during an obstacle avoidance task were carried out. The implemented neural model was introduced by Izhikevich (2003); this model is capable of reproducing abrupt changes in the membrane potential of biological neurons, known as action potentials. The learning strategy is based on a multimodal association process where the synaptic weights of the networks are modified using a Hebbian rule. Despite the growing interest generated by artificial action potential neural networks, there is little research that implements these models for learning and the control of autonomous agents. The present work aims to fill this gap in the literature and at the same time, serve as a guideline for the design of further experiments for in vitro experiments where neural cultures are used for robot control.

Too many research works have focused on the problem of control of robot manipulators while executing tasks that do not involve the contact forces of the end-effector with the environment. However, many tasks require an interaction of the manipulator with the objects around it. For the correct performance of these tasks, the use of a force controller is essential. Generally, the control objective during the contact is to regulate the force and torque of the manipulator’s end-effector over the environment, while simultaneously regulating the position and orientation (i.e., the pose) free coordinates of the manipulator’s end-effector. Many works have been presented on this topic, in which various control strategies are presented; one of the most relevant methods is the so-called hybrid force/position control; this scheme has the advantage of being able to independently control the force in constrained directions by the environment and the pose along unconstrained directions. This work analyzes and implements the hybrid force/position control using a fuzzy logic control method, since the fuzzy control provides a solution for nonlinearities, high coupling, and variations or perturbations. The system employed is the Mitsubishi PA10-7CE robot manipulator, which is a robot of 7 degrees of freedom (DOF), but in this work, it is only used as a 6-DOF manipulator, equipped with a 6-DOF force/torque sensor in the end-effector.

In recent years, finite-time controllers have attracted attention from some researchers in control, who have formulated applications to several processes and systems, including serial robotic manipulators. In this work, we report the application of a finite-time nonlinear PID controller to a Five-Bar Mechanism, which is a parallel manipulator, for set-point controller. The stability analysis of the closed-loop system shows global finite-time stability of the system. The dynamic model of the Five-Bar Mechanism developed in this work is a so-called reduced model, which has a structure similar to a serial robot. Moreover, the results of the numerical simulations carried out confirm the usefulness of the proposed application. The contribution of this work is to show the feasibility of the application of a finite-time nonlinear controller to a Five-Bar Mechanism and the usefulness of the proposed approach by numerical simulations.

This chapter deals with the tracking control problem of a three-degree-of-freedom (3-DOF) helicopter. The system dynamics are given by a mathematical model that considers the existence of a dead-zone phenomenon in the actuators, as well as a first-order dynamic that adds a lag in the system input. This leads to obtain an eighth-order model where the positions are the only available measurements of the system. The control problem is solved using nonlinear \( {\mathcal{H}}_{\infty } \) synthesis of time-varying systems, the dead-zone is compensated using its inverse model, and a reference model is used to deal with the first-order dynamic in the actuators. Numerical results show the effectiveness of the proposed method, which also considers external perturbations and parametric variations.

People who have suffered an injury require a rehabilitation process of the affected muscle. Rehabilitation machines have been proposed to recover and strengthen the affected muscle. In this chapter, we propose a novel ankle rehabilitation parallel robot of two degrees of freedom consisting of two linear guides. For the integral rehabilitation, a serious game and a facial expression recognition system are added for entertainment and to improve patient engagement in the rehabilitation process. The serious game has a simple design. This game has three levels and it is controlled with an impedance control, which specific command allowing character game jumps the obstacles. Facial expressions recognition system assists to the serious game. We propose to recognize three different facial expressions to the basic expressions. Based on the experiment results, we concluded that our system is good because it has a performance of 0.95%.

Recent technological advances have provided the manufacturing industry with precise and robust machines that perform better than their human counterparts in tiresome and tedious jobs. Likewise, robots can perform high precision tasks including in hazardous environments. However, a new area of research in robotics has emerged in the last decades, namely cognitive robotics. The main interest in this area is the study of cognitive processes in humans and their implementation and modeling in artificial agents. In cognitive robotics, the use of robots as platforms, in the study of cognition, is the best-suited mechanism as they naturally interact with their environment and learn through this interaction. Following these ideas, in these works, two low-level cognitive tasks are modeled and implemented in an artificial agent. Based on the ecological framework of perception, in the first experiment, an agent learns its body map. In the second experiment, the agent acquires a distance-to-obstacles concept. The agent is let to interact with its environment and allowed to build multimodal representations of its surroundings, known as affordances. Internal models are proposed as a conceptual mechanism which performs associations between different modalities. The results presented here provide the basis for further research on the capabilities of internal models as a constituent cognitive base for higher capabilities in artificial agents.

Haptic systems and virtual environments represent two innovative technologies that have been attractive for the development of applications where the immersion of the user is the main concern. This chapter presents a brief review about applications of haptic systems in virtual environments. Virtual environments will be considered either virtual reality (VR) or augmented reality (AR) by their virtual nature. Even if AR is usually considered an extension of VR, since most of the augmentations of reality are computer graphics, the nature of AR is also virtual and will be taken as a virtual environment. The applications are divided in two main categories, training and assistance. Each category has subsections for the use of haptic systems in virtual environments in education, medicine, and industry. Finally, an alternative category of entertainment is also discussed. Some representative research on each area of application is described to analyze and to discuss which are the trends and challenges related to the applications of haptic systems in virtual environments.

This paper presents the aerodynamic analysis of a bio-inspired empennage, which mimics the way that the tail of some birds moves. To avoid a kinematic chain in the proposed design, the three axes of rotation intersect in a single point. To define the relationship between the attitude of the empennage and the aerodynamic coefficients, different tests were conducted in an open-circuit wind tunnel at low velocity, where the attitude of the empennage was changed, and the aerodynamic effects were measured by means of a force sensor and expressed with six dimensional wrenches, including forces and torques. Wrenches were transformed from the sensor’s reference frame to a frame located at the aerodynamic center of the empennage to compute the aerodynamic coefficients. A multiple regression analysis revealed a coupling effect between the aerodynamic coefficients and the attitude of the proposed empennage, as changes in any of the three Euler angles modify the aerodynamic coefficients. Besides, it is shown that both the longitudinal and the translational motion of the vehicle can be controlled by the proposed bio-inspired empennage.

In this article, the problem of performing different tasks with a group of mobile robots is addressed. To cope with issues like regulation to a point or trajectory tracking, a consensus scheme is considered. Three topologies were tested in simulation. The first goal was to make consensus in the group of robots, after the consensus point was relocated to achieve a regulation control. The last objective was to follow a desired trajectory moving the consensus point along the predefined path. The proposal was validated through experimental test with a group of three differential mobile robots.