previous day
next day
all days

View: session overviewtalk overview

09:00-10:00 Session 5: Keynote Lecture
Key outcomes of the Task Force on Definition and Classification of Power System Stability – Revisited & Extended
10:10-11:50 Session 6A: Internet of Things
Embedded system for agriculture variables monitoring based on IoT technology.

ABSTRACT. This paper defines the construction and implementation of an embedded system through advanced information and communication technologies in agriculture (smart and precision agriculture). The development consists of an embedded system that integrates a GPS module with sensors for monitoring agricultural variables such as the chlorophyll index in plants, pH, electrical conductivity, and soil temperature. In addition, this device integrates a solar panel with a battery through a control system. The precision of this device was contrasted with measurement instruments used in soil laboratories, achieving adequate performance for monitoring agricultural variables that facilitate decision-making in crop management. Another aspect to highlight is the robust and modular design, developed and implemented in the casing and structural components of the embedded system.

A Design of a Physiological Parameters Monitoring System, Implementing IoT Communication Protocols by Using Embedded Systems.

ABSTRACT. In this paper the procedure and methodology for the wireless communication of biomedical sensors is shown. In this case of study, we focused on Photoplethysmography (PPG) Heart Beat Rate and Respiratory Rate sensors. This communication system was made using the ESP32 and Beagle Bone Black embedded systems. Also, a graphical user interface (GUI) and data storage system was developed in order to facilitate its visualization, analysis and consequentially the decision making.

Subscriber Location in 5G mmWave networks - Machine Learning RF Pattern Matching

ABSTRACT. A realistic simulated 5G DM-MIMO wireless network operating at 28 GHz mmWaves has been deployed using Open Street Maps and Matlab® over the campus of Universidad San Francisco de Quito (USFQ). Received Signal Strength fingerprints have been collected at Base Station antenna array, and the K-Nearest Neighbors method has been used to perform the match between the received RF patterns and the stored fingerprints. Three different procedures were tested and their results were compared, exhibiting very good outcomes in all the cases.

Natural Language Processing: A proposal for Word-per-Minute Evaluation in Students’ Performance within the Classroom

ABSTRACT. The objective of this research is to evaluate the number of words per minute by using a mobile application that implements the natural language processing and that contributes to the reading comprehension of students of 5th and 6th grade of elementary education in Mexico. The application was developed with the software development kit Flutter to do the interface programming and the library “Vosk” that implements the natural language processing. The results demonstrate the ability of the mobile application to detect the words that student is capable to read during the aloud reading. In addition, the application registers the time of the reading and finally, ask questions at the end of each proof. These results allow to calculate the reading comprehension of the student by the natural language processing that give a general result of the performance at the end of the test.

A Cloud Remote Monitoring System of Hydraulic Pressure by a Minicomputer

ABSTRACT. In this work, a cloud remote monitoring system to sense the hydraulic pressure of a piping is implemented on a single-board computer, Raspberry Pi. Furthermore, an alert scheme is designed to send by e-mail messages of warnings when the measured variable is out of range. In this manner, operators could be able to know the status of a hydraulic process even if they are situated far away from the physical equipment. The monitoring system includes the setting up and characterization of an analog pressure sensor, which requires the use of an analog-digital converter to be connected to Raspberry Pi. By means of Python programming language, the data acquisition and cloud computing are achieved with the support of the ThinkSpeak platform. Experimental probes show a feasible performance about to use a single-board computer to applications of the Industrial Internet of Things.

10:10-11:50 Session 6B: Power Converters Applied to Renewable Energy Systems
Simultaneous MPPT and MEPT for Asymetric Parallel Boost-Type Converters

ABSTRACT. Energy harvesting from different sources is challenging if maximum power and efficiency must be tracked in green energy intermittency scenarios. Studies for simultaneous maximum power and efficiency points tracking are aimed at a single source of considerable power generation capability. This paper proposes a low computational burden algorithm for simultaneous maximum power and efficiency point tracking for a paralleled-input single-output boost-type electronic converter operating in DCM with sequential triggering, asymmetrical inputs, and low-power sources. In numerical simulations, a stable 95% of efficiency is gained over a wide range of operation conditions, even considering noise (bounded but reasonable) and parasitics of the main components.

Control Strategy for a Dedicated On-Board Isolated Battery Charger

ABSTRACT. This paper presents the design and modeling of a controller applied to a battery charger. The battery charger topology is composed of two stages, namely the PFC stage and the isolated DC-DC power supply. The first one consists of a boost-rectifier stage connected to the grid based on Full-bridge topology; the second stage is an isolated DC-DC power supply associated with the battery charging. The last is composed of three main components, an inverter that generates high-frequency AC to produce a magnetic field of a transformer; the transformer provides galvanic isolation to the battery side, and a bridge rectifier connected to the batteries. Both stages are linked through a capacitor that functions as the output of the PFC boost-rectifier stage and as a constant input source to the DC conversion stage. The control strategy is composed of three control loops, referred to as, the AC current-control loop that forces the grid current to track the desired reference which is constructed as a signal in phase to the sinusoidal input voltage, a DC voltage-mode control loop that is aimed to regulate the DC capacitor voltage, and finally, a battery charging control-loop designed to assure proper battery charging under the constant-current and constant-voltage charging profile. Besides, adaptive estimators are considered to deal with parametric uncertainties of the input impedance. Finally, numerical results are obtained to evaluate the proposed control law.

Real-Time Simulation of Model Predictive Control for Boost Converters in the Linear Complementarity Framework

ABSTRACT. This paper presents the use of the linear complementarity (LC) framework for power converter modeling, where non-linear characteristics of the converters are modeled (switched control signals and the conduction modes; continuous and discontinuous). The LC model is used in a model predictive control (MPC) scheme with a pulse-width modulation, where the MPC horizon is a duty cycle period. The MPC is solved by a non-linear optimization problem off-line, for a finite values of current and voltage through parametric analysis, using the Nelder-Mead algorithm and the PATH solver. In this work, a parametric analysis is proposed by fitting the controller optimal duty cycle response through the general equation of the plane. The MPC is implemented in a microcontroller and applied to a boost converter, where the performance is tested in real-time using a Hardware-in-the-Loop (HIL) method.

Solid State Transformers Bi-directional control based on Dual Active Bridge

ABSTRACT. The  electrical distribution network is continuously changing due the fact that in recent years the distributed generation sources interacting with it have increased, this impacts direcly the energy consumption dynamics because the network does not just deliver energy but can also receive it. These new operating conditions requires new and modern conditioning stages and power flow control rules to ensure optimal system performance. One of the alternatives currently being studied and developed to face this challenge are Solid State Transformers (SST). These devices are composed by Power Electronic Converters and have diverse attractive features which make them a vital, world-class study case; one of his main requirements its to ensure a proper operation on the SST power flux direction.

This paper describes a control proposal applied to a Dual Active Bridge (DAB) converter, which is part of an SST; the proposal serves as the starting point to establish new control rules that satisfy the basics criteria that a bidirectional converter needs to meet.

Adaptive IDA-PBC for Output Voltage Regulation of a Fuel Cell Hybrid Storage System

ABSTRACT. In this paper, based on interconnection and damping assignment passivity-based control approach, a multi-loop adaptive controller for output voltage regulation of a hybrid system comprised of a proton exchange membrane fuel cell energy generation system and a hybrid energy storage system consisting of supercapacitors and batteries, is detailed. The control scheme relies on the design of two control loops, i.e., an outer loop for voltage regulation through current references generation, and an inner loop for current reference tracking through duty cycle generation. Furthermore, an adaptive law based on immersion and invariance theory is designed to enhance the outer loop behavior through unknown load approximation. Finally, numeric simulation results show the correct performance of the adaptive multi-loop controller when sudden load changes are induced in the system.

10:10-11:50 Session 6C: Vision and Language
Sequential Models for Automatic Personality Recognition from Multimodal Information in Social Interactions

ABSTRACT. The task of automatic personality recognition has become very popular in recent years and it is considered a difficult one as we are trying to model human behavior that may not be visually obvious. Although state-of-the-art approaches have used deep learning architectures such as Transformers and some techniques such as Neural Architecture Search (NAS), some of these methods disregard valuable temporal information. In this paper, we approach the task by modeling it as a sequential problem, using a bimodal recurrent neural network, and exploiting the visual and textual modalities jointly. We report experimental results obtained in a novel corpus of dyadic interactions, outperforming state-of-the-art for the Extraversion personality trait. Another contribution of this paper is that we also analyze the regression to the mean problem that we think most state-of-the-art approaches could be facing when approaching the personality recognition task.

2D automatic positioning system by means of a PID control in optical spatial filtering devices

ABSTRACT. In this work, an interface that allows the automation of the alignment process in spatial filter systems is proposed. Optical spatial filtering is a method used in interferometry for the generation of regular laser illumination beams, where a pinhole aperture is focused on a convergence point of a lens. For the experimental setup, an emerging laser beam is focused to a point, where the correct position of the aperture can be found by adjusting the linear displacement with motorized devices. To obtain the position of the pinhole in an (X,Y) plane, a displacement arrangement implemented with micrometric servomotors and a CCD camera; controlled by a PID control algorithm, is used. The automatic adjustment of the position of the resulting beam intensity pattern registered in the camera is carried out by the servo motors until the optimal position coordinates are found. Process control related to positioning and subsequent estimation is executed with Python and synchronized in LabView.

Watermarks based on Pyramidal Images for Tampering Image Self-Restoration

ABSTRACT. This article presents the Pyramidal Image Watermarking (PIW) and Pyramidal Image Restoration (PIR) algorithms based on a pyramidal image representation. The PIW algorithm watermark a digital image with an array of compressed images corresponding to a pyramidal representation and the PIR algorithm is capable of performing automatic image restoration using the watermark images computed by the PIW algorithm. The use of pyramids avoids the quantization procedure for the reference information bits and allows us to identify pixels erroneously deemed as altered by the block-wise tampering detection scheme. PIW and PIR outperform some state-of-the-art algorithms used for automatic image restoration and can reconstruct images with a tampering rate of up to 0.95.

Watermarking Scheme With Bounded-Exhaustive Self-Recovery Approach

ABSTRACT. This paper revisits a fragile watermarking scheme with exact self-recovery capabilities and proposes a novel restoration process. The restoration mechanism is aimed at solving systems of binary linear equations (SBLEs) associated with altered subsets of bits, using arithmetic module-2. The chances that those SBLEs will be linearly dependent increase for more significant tampering rates. While such subsets are deemed to be unrecoverable by schemes in the current watermarking literature, the proposed restoration process calculates the Hermite standard form of the matrix representation of the SBLEs to find both the pivot and the free variables. Since the number of solutions is finite when using arithmetic module-2, all the answers are calculated by exhaustively replacing the free variables with all the possible combinations of bits. Finally, the solutions are analyzed to possibly restore some tampered bits to enhance the recovery performance in every iteration. Experimental results demonstrate that the proposed scheme outperforms the best methods with exact self-recovery capabilities in the state-of-the-art in terms of embedding time, restoration time, and restoration performance.

Automatic Selection of White Paint Types for Automotive Industry

ABSTRACT. In this work, we design an optical experimental setup and a data augmentation methodology to generate 384 images of automotive parts, painted with two different types of white paint. We also perform the automatic classification of the database considering two different classes. To obtain the highest precision, we have used two different classification scenarios, 3 algorithms, and 4 metrics. Also, we assume that the results can be improved by extracting the image characteristics using the convolutional neural network ResNet50 and using them as an input. Our results show that an error-free classification can be obtained independently of the scenario or classifier. We obtained 100 % in each of the 4 metrics in the six studied variations. Therefore, the machine time is the parameter we can use to select the optimal classifier, where the classifier Multinomial Naïve-Bayes under the Training and Test Set scenario was the fastest algorithm with 65 s, and the classifier Machine Vector Support under the Cross-Validation scenario, the slowest with 79.12 s.

10:10-11:50 Session 6D: Renewable Energy
Impact of high penetration of photovoltaic distributed generation on the protection coordination in distribution networks

ABSTRACT. Nowadays, it is increasingly common to see an increase in the interconnection of Distributed Generation Systems (DGS), mainly using photovoltaic modules, in the Distribution Networks (DN). This is because of the reduction of the cost to buy these generation systems. The continuous increase of DGS brings some technical challenges that must be addressed. One of these technical challenges occurs in the DN's protection schemes since a high concentration of Photovoltaic Distributed Generation Systems (PDGS) can cause changes in the operating times of the protection devices and therefore cause a loss of coordination between them. Thus, to comply with the change and operating criteria of the DN protections, the change caused by a large concentration of PDGS must be considered. This paper evaluates the effect on the DN protection schemes caused by different PDGS penetration scenarios. It is considered in the analysis the technical restrictions required to protect a DN, as well as the limits to interconnect PDGS. To evaluate this impact, the DIgSILENT PowerFactory software and the IEEE 13 Node Test Feeder are used. The obtained results show how the increasing penetration of PDGS in the DN affects its protection schemes.

Optimal Sizing of a Stand-alone Renewable-Powered Hydrogen Fueling Station

ABSTRACT. In this paper, we propose a model for optimal sizing of the key components of a renewable-powered hydrogen production and fueling station. Renewable energy generated from on-site wind and solar resources are used to generate hydrogen using electrolysis. The generated hydrogen is stored in an on-site hydrogen storage tank and used to fuel a total hydrogen demand of two ton per day. The model is based on a stochastic mixed-integer linear programming formulation that solves for optimal sizing of the wind turbines, the photo-voltaic arrays, the hydrogen production capacity of the electrolyzers, and the storage capacity of the hydrogen tank. Numerical results are provided using available cost parameters in the context of Canadian market. The simulation results show that a near-green hydrogen fueling station powered by only wind and solar energy could produce hydrogen at under 6.8 $/kg when financing costs are considered or under 5 $/kg when financing costs are neglected. A hybrid ‘turquoise’ fueling station that supplies hydrogen using a mix of generated green hydrogen and imported blue hydrogen (<18 %), could produce hydrogen at under 5 $/kg considering financing cost or under 3 $/kg when such costs can be neglected.

Minimization of energy storage in a solar farm by controlling the state of charge of a battery bank

ABSTRACT. This paper presents a methodology for reducing energy storage requirements by using batteries to smooth out power variations of a solar farm. This reduction is achieved by combining the operation of an energy management system (EMS) with an optimally sized battery energy storage system (BESS). The capacity calculation of the BESS is determined by using an evolutionary algorithm. The EMS deploys a state-of-charge feedback (SOC-FB) control system to reduce the power variations of a solar farm, while the optimization algorithm determines a minimum capacity of the BESS to keep the power variation under the specification of a cost function. This approach is tested by using real data from a small solar farm. Simulation results show a significant reduction of the power variation of the solar farm output, with a minimum capacity of the BESS.

Performance analysis of a novel Zinc-air battery powering an IoT node

ABSTRACT. — In this work, the performance of a Zinc-air battery with a novel structure powering a Bluetooth IoT node is described. Although these batteries have the highest energy density of all available battery technologies, commercial Zinc-air batteries are not suitable for wireless nodes because they are not capable of enduring the short high-current pulses that occur when radio operation is active. Results show that the properties of the battery under test are superior to standard commercial zinc-air batteries and competitive with Li-ion batteries performance, making it suitable for IoT applications.

Design of a didactic bench of low-power wind generators with Savonius turbines

ABSTRACT. On this paper, a Savonius Wind Turbine prototype is proposed as a non-centralized option for low-scale power generation. Recent studies about Savonius wind generators have been developed in last years, as they are an attractive alternative to conventional horizontal axis wind turbine. The main advantage of vertical axis wind turbines is that they don’t require additional wind orientation mechanisms, avoiding complexity of design related to stresses generated in the turbine vanes in events of abrupt changes in wind orientation. The methodology carried out in the manufacture of the prototype in the present work is based on the development of the mechanical design using the SolidWorks software, moving on to the calculations of the estimation of electrical power, that after a validation of those calculations with the design. The dimensions of the principal elements of the prototype are next: the blades are half section of circumference with diameter of 0.25 m and the endplates of the turbines have 0.5 m of diameter. To be able to estimate the wind potential of a region, the method uses was a probability distribution that accurately describes the frequency (or probability) with which a specific value of wind speed occurs, called the Weibull distribution. The average speed in the region of Hermosillo, Sonora, based on a Global wind atlas, is 2.42 m/s. Considering the calculation of the weighted average power, the generated result is 15.98 W/m2, compared to the power generated only with at an average wind speed will is of 8.64 W/m2. The results showing the importance of carrying out the study with the weighted average power.

10:10-11:50 Session 6E: Industrial
Simulación en Tiempo Real
Viakon y Prolec GE, cables y transformadores eléctricos
Oferta de Maestría y Doctorado en Ciencias en Ingeniería Eléctrica
10:10-11:50 Session 6F: Industrial
EconiQ - Portafolio ecológico para un futuro energético neutro en carbono
Gemelos digitales para electrónica de potencia
Programas del Posgrado en Ciencias en Ingeniería Electrónica
12:10-13:50 Session 7A: Electrical
DC line current in a compact MTDC system for DC short-circuit fault analysis

ABSTRACT. In the search for alternatives to solve the challenges that protection philosophy entails in MTDC systems due to their low resistance, it is desirable to have a base system where it is possible to analyze the various types of short-circuit faults, the contribution of fault current from other transmission lines and the contribution of fault current from the AC system through the stations. Currently, most MTDC systems available through transient analysis tools include complex voltage, power, and frequency controls, to name a few. Since these controls do not have a significant influence on the amplitude and behavior of a short-circuit fault current, they can be eliminated to reduce the computational load. In this way, the fidelity of the behavior of the fault current increases and leads to a precise analysis of the short-circuit phenomenon that opens up new proposals for fault detection and mitigation schemes. In order to clearly analyze the different types of faults, a compact CIGRE four station MTDC system is proposed through the PSCAD/EMTP transient analysis tool.

Unit Commitment Analysis in Hybrid Networks Containing Point-to-Point VSC-HVDC links

ABSTRACT. This paper presents an efficient modeling approach for solving the Unit Commitment (UC) problem in hybrid power grids. This is based on the linear modeling of the AC network and point-to-point High Voltage Direct Current (VSC-HVDC) links through a nodal power flow approach using AC voltage phase angles and DC voltages. This results in a Mixed–Integer Linear Programming (MILP) formulation. The obtained results permit to determine the optimal hourly scheduling for VSC-based transmission grids in the context of day-ahead electricity markets. The applicability of this formulation is showcased using two empirical power systems incorporating a point-to-point VSC-HVDC link.

Shooting-Method-Based Simulink Initialization

ABSTRACT. This work elaborates the implementation and performance of a shooting method for rapid initialization of electrical power systems models implemented with specialized power systems libraries of Simulink. We selected the numerical differentiation method as the shooting method since it does not need the mathematical model of the dynamic system explicitly, but only the state vector at the beginning and the end of the fundamental period. Additionally to the initialization, the method provides stability of the solution through the Floquet multipliers. We tested the method with a STATCOM based on a two-level inverter and the results exhibit its correct performance in accuracy and computation time.

Electromagnetic-thermal analysis of core fastening elements in a three phase reactor

ABSTRACT. In this article, the temperature distribution in one of the core fastening bolts of a three-phase shunt reactor under fault and non-fault conditions is obtained by applying the finite element method (FEM). A preliminary model is proposed that describes a failure that occurred progressively in a reactor due to the fact that the insulation of the bolt was damaged and caused an undesired temperature increase. For the analysis, a three-dimensional (3D) finite element model is used to obtain the distribution of the electromagnetic field, as well as the temperature in the area where the reactor failure occurred.

Short-circuit impedance calculation in a power autotransformer using the finite element method

ABSTRACT. In this article, the percentage of short-circuit impedance is calculated using the finite element method for a single-phase power autotransformer with tertiary winding of 25 MVA, 230 kV, 60 Hz shell type of three columns using the B-H curve corresponding to the core material. A 3D model and the formulation of the magnetic vector potential are used. This type of study is essential to guarantee the safety and operability of electrical power systems. To calculate the percentage of short-circuit impedance , the classic method is used using the voltage relationship between the short-circuit voltage and the nominal voltage, in addition, the magnetic energy storage technique is used too to calculate this parameter. The results are validated with the data from the laboratory tests obtaining differences of less than 5 %.

12:10-13:50 Session 7B: Power Converters Applied to Renewable Energy Systems
Switching Analysis in a Back-to-Back converter of the Type-4 Wind Turbine

ABSTRACT. In this paper a wind energy conversion system (WECS) based on the type-4 topology is shown. The system includes a permanent magnet synchronous generator (PMSG) and a Back-To-Back converter controlled by PI controllers. Average, approximate pulse width modulation (PWM), and ideal PWM switching are applied to the Back-To-Back converter's voltage source converters (VSCs). The approximation model is focused on the hyperbolic tangent (HT) function. Fourier series is implemented to show the harmonic spectrum caused by switchings. In addition, the effect of the error caused by the increase of the integration step and reduction of simulation times is shown.

Analysis and Design of Buck Converter R2P2 with Interleaved Function

ABSTRACT. The wide variety of applications that require power conversion apply DC/DC converters with wide transformation ranges and the capacities to provide high current or voltage levels. Low voltage applications have been standardized to 48V, 24V, 12V and 5V values for different applications such as LED lighting, battery banks and power supplies for telecommunication system. Some requirements that DC/DC converters must meet are high transformation ratios, high power density, high efficiency and low current and voltage ripple at the input and output. In this work, the analysis and design of a Quadratic Buck Converter based on the concept of R2P2 with interleaved function is proposed for low voltage and high current applications. The operating modes of the converter during the switching process of the switches are presented. The average models are also presented as well as some comparative aspects with some topologies reported in the literature.

High Gain Step-up DC-DC Converter With Low Duty Cycle Operation

ABSTRACT. In this paper, a dc-dc switching high step-up converter with reduced duty cycle operation is presented. The converter is based on reduced redundant power processing (R2P2) concept, using a noncascading structure I-IIA. The analysis of the converter is carried out with different sources and loads to verify the differences in a typical and photovoltaic application. In this work, the voltage conversion ration, steady state operation and semiconductor stress of the proposed converter are discussed.

12:10-13:50 Session 7C: Artificial Intelligence in Electrical Systems
An Efficient Neurocontroller Position Method for PMSM Drive System

ABSTRACT. PMSM has been widely used in high-precision variable-speed applications, however, the control scheme demands normally a high dynamic performance under several operating contidions. Due to the non-linear nature of the PMSM, the use of an adaptive controller based on B-spline neural networks is proposed to determine the control signals. The proposed control technique through neural networks exhibits the best performance because it can be adapted to each operating condition, demanding low computational cost for an online operation, and considering non-linearities of the system. The performance of the proposed controller is evaluated in the presence of uncertainties. The results are compared with the conventional PI controller, optimized using whale optimization algorithm.

Non-linear PID Controller Optimization using the Artificial Bee Colony Algorithm Applied to a Small-Scale Pasteurization Plant

ABSTRACT. This paper investigates the performance of the artificial bee colony (ABC) algorithm as an alternative to finding the best optimal gain parameters for a nonlinear PID (NLPID) controller applied to the heat control in a nonlinear pasteurization process. Comparisons between a traditional PID and the NLPID optimized by ABC were carried out in terms of setpoint step changes. The experimental results reveal that the ABC-based tuned NLPID controller provides better tracking of the nonlinear dynamical system concerning convergence time, perturbation rejection, and a smoother and shorter duration of the control action. Therefore, these results demonstrate that the ABC algorithm can be used as an alternative to facilitate the tuning of the nonlinear parameters of NLPID controllers.

Optimal Switching Angles Calculation for a Multilevel Inverter Through the ABC Algorithm

ABSTRACT. This paper explores the use of the artificial bee colony algorithm to obtain the switching angles of a five-level Cascaded H-bridge Multilevel Inverter. The objective of the algorithm is to reduce the Total Harmonic Distortion in the inverter output voltage. These switching angles are compared in a laboratory prototype with the fsolve method, which is included in the Matlab software. The results show a better Total Harmonic Distortion of the output voltage compared with the fsolve method.

12:10-13:50 Session 7D: Computer / Biomedical Applications
A Proxemic Potential Field Approach for Modeling Interactions between Autonomous Vehicles with Pedestrians and Cyclists

ABSTRACT. Autonomous vehicles (AVs) are designed to drive in urban areas with a high volume of people on the street. In these locations, the safety of people around and inside the vehicle relies on its capacity to react to possible collisions. Typically, the motion of the obstacles involved is static or follows a well-known trajectory, but the people instead can change freely their movement with time, therefore, AVs should consider them as dynamic obstacles. In this work a proxemics based model is developed to represent in real time the interaction zones, between AVs and pedestrians or cyclists, through elliptic regions depending on instantaneous motion parameters. The model was implemented in simulations involving pedestrians and cyclists walking in street-based scenarios. The results suggest that the proposal allows to react and break a vehicle in response to a possible collision.

Visual Sensor System Applied to Trajectory Generation for UAVs

ABSTRACT. In this work, a visual sensor system is applied to generate free collision trajectories using a new algorithm called Trajectory Generation Based on Remote Object Detection (TGBROD). The TGBROD algorithm is based on remote object detection through a video received from a drone to a station work. The video analysis is done by the TGBROD algorithm, and as a result, free and collision regions are defined. Using information on the regions and the drone position, free and collision vector trajectories have been generated. The trajectories were mathematically represented as the vector, parametric and symmetric equations. The free vectors find practical applications for UAVs since these can be used as references for the drone controller.

Exposure Time and Depth Effect in Laser Speckle Contrast Images under an Adaptive Processing

ABSTRACT. Blood vessel visualization aims to find structures related to microvasculature within the biological tissue and has been applied to diagnose and treat vascular diseases based on vessel size in ophthalmology, dermatology, and neuroscience, among others. Laser Speckle Contrast Imaging (LSCI) is a technique developed to estimate the relative blood flow speed and to improve blood vessel visualization. However, the highly scattering of the surrounding tissue hinders the blood vessel visualization and introduces an intense noise level proportional to the depth of the blood vessel perceived as spatial variations in the image; under these conditions, standard LSCI is limited. In this work, in order to calculate speckle contrast, we applied adaptive processing to raw speckle images acquired from a skin phantom varying the exposure time and blood vessel depth. Experimental results show that it is possible to improve the quality of CIs through adaptive processing for superficial and deep blood vessels.

Robotic Arm Handling Based on Real-time Gender Recognition Using Convolutional Neural Networks

ABSTRACT. This paper presents the development of a system for controlling a robotic arm to deliver an object depending on the gender identity (male or female) of a human recognized in front of the robot. For this, we developed a convolutional neural network-based model for recognizing genders. With the recognition result, the control of the robotic arm with six degrees of freedom was implemented using a Jetson Nano embedded computer, OpenCV, ROS, and TensorFlow libraries. The developed gender identification model achieved a 96.5% of accuracy and a loss of 3.5\% during training and validation using a gender database composed of 50K gender images. While the final real-time prototype obtained a 98.2% accuracy and a margin of error of 1.8% during testing.

14:10-15:10 Session 8: Keynote Lecture
Synchronized-Measurement Applications in Power System Stability, Protection and Control