Proceedings of Congress on Control, Robotics, and Mechatronics

Increased shortage of skilled workers and increased demand for goods tend to strain manufacturing activity. This issue, along with a poor, hazardous working environment, furthers the need to robotize the activity. Welding, one of the major manufacturing processes, has witnessed automation in the last two decades. Welding using robots is mainly accomplished in ‘teach and playback’ mode. It necessitates reconfiguration every time the robot engages in a new task. Knowing the weld joint beforehand allows the programmer to set relevant parameters in advance. Hence, this study aims to solve the issue by proposing an alternate way to automatically recognize weld joint types. This paper suggests an effective way to classify the weld joint type using the image processing and feature extraction technique. The method works in two stages: features extraction and bag of visual words (BoVW) model building. First, image processing algorithms are used to condition the greyscale image. Image conditioning involves noise removal using a contrast-limited adaptive histogram equalization (CLAHE) and enhancement to improve the image’s contrast. Then SURF features of processed images are extracted and input into a support vector machine (SVM)-based bag of visual words classifier for classification. The method is capable of recognizing five types of weld joints. The bag of features strategy combined with SVM yields 97% accuracy.

A person with a mental disorder exhibits a significant disturbance in his or her behavior. Generally, mental disorders are associated with distress or impairment of normal functioning. Lack of adequate resources and facilities, as well as a lack of awareness of the symptoms of mental illness, prevent people from getting the help they need. The ability to assess depression through speech is a critical factor in improving the diagnosis and treatment of depression. The spoken language is said to provide access to the mind, and a wide range of speech capture and processing technologies can be used to analyze mental health. Speech processing is about recognizing spoken words. The automatic recognition and extraction of information from speech enables the determination of some physiological characteristics that make a speaker unique to identify their mental health status. In this paper, we describe how mental health-related problems can be predicted by speech processing. This paper identifies the gaps in the literature review that lead to the proposed methodology.

Reducing power loss and meeting escalating load requirements are one of the most important goals in the distribution system. This paper discusses ways to optimize placement and sizing of distributed generators (DGs) to reduce power loss and meet increasing load requirements in the most efficient way, thereby supplying clean energy. The best size of distributed generating unit is found through a genetic algorithm inclusive of minimizing the losses in real power, and location is determined based on minimum real power loss and improving the stability index. The two goals of power loss reduction and stability are in conflict with each other. So, an optimal solution is achieved which meets both objectives. The genetic algorithm is employed to search out a group of optimal solutions satisfying those two objectives. To solve the extremely nonlinear issue of computing the total power loss under operational equality and inequality requirements, the genetic algorithm (GA) is applied. A simulation-based analysis is performed on the IEEE-14 bus system to verify the simulation results using Simulink.

High-performance fibre-reinforced cementitious composite (HPFRCC) has been demonstrated to provide superior tensile ductility and fracture energy compared to normal concrete at both quasi-static and dynamic strain rates. For this reason, this material becomes potential material for application to structures subjected to dynamic loading. However, there is still a lack of accuracy model for estimating strain-rate dependence of tensile ductility of HPFRCCs since most current empirical regression models have been proposed based on individual limited test data. In this study, a model-based neural network has been trained to estimate the strain-rate dependence of tensile ductility of HPFRCCs using 150 tensile test results. There are six input variables: matrix strength, fibre type, fibre length, fibre diameter, and fibre volume content, while strain-rate dependence of tensile ductility is output parameter. The results of prediction showed that the machine learning-based model was an efficient method to estimate strain-rate sensitivity in tensile ductility of HPFRCCs with high accuracy. By performing sensitivity analysis, the relative importance of all influencing factors was determined.

Machine learning is used to analyze data from different perspectives, summarize it into useful information, and use that information to predict the likelihood of future events. Classification is one of the main problems in the field of machine learning. The aim here is to study various classification algorithms in machine learning applied on different kinds of datasets. The algorithms used for this analysis are J48, Naive Bayes, multilayer perceptron, and ZeroR. The performance is analyzed using various metrics such as true positive rate, false positive rate, and error rates such as root mean squared error and mean absolute error. The performance of J48 algorithm is better than other algorithms for large datasets. The proposed algorithm still increases the performance in terms of error rates for large datasets. The contemplated algorithm is eventuated by mutating the splitting paradigm in the tree-based algorithms. The experimental analysis demonstrates that the proposed algorithm has reduced error rate as compared with the traditional J48 algorithm.

The development of industrial and service robots in recent years has attracted a lot of attention to robotics research. Commercially available robotic grippers are sometimes costly and difficult to customize for individual applications. Therefore, an open-source, low-cost three-finger robotic gripper has been introduced in this paper. The main focus was to create a 3D-printed model. 3D printing technology solves the issues of cost and weight for the implementation of designs. Linear-bearing LM8UU hard-chrome smooth rods were used in the design. Flexible couplings were used to provide motion to the fingers of the gripper. DC servomotor was used as an actuator in the model. The proposed system used Raspberry Pi as the processor and L298N as the motor driver. The gripper was designed to provide the ability to perform grasping a variety of household objects. The gripper was tested on 7 household objects which included a square box, cylindrical object, broom, cup mug, bag, and bottle. The gripper performed well while grasping different objects. The gripper was successful in grabbing objects of various shapes and sizes to the weight of 1000 gm.

High-performance fiber-reinforced cementitious composite (HPFRCC), a new class of concrete technology, exhibits outstanding mechanical resistance, especially in terms of superior post cracking strength, strain capacity, and energy absorption capacity. Among mechanical properties, energy absorption capacity of HPFRCCs has become one of the most popular properties that received much attention from researchers to discover and model. However, a more accurate model for prediction of energy absorption capacity is still discouraged to develop since current empirical regression models based on limited data have shown their limitations. In this research, the energy absorption capacity of HPFRCCs is predicted through a proposed machine learning-based model using 103 tensile test results. The input variables include matrix strength, fiber type, fiber length, fiber diameter, and fiber volume content, while the output variable consists of energy absorption capacity. From the prediction results, the energy absorption capacity could be predicted well using machine learning based models. From the results of sensitivity analysis, the contribution of each input variable to the energy absorption capacity of HPFRCCs was figured out.

Online movie ratings have evolved into a serious business. Hollywood generates around $10 billion in box office revenue in the United States each year, and online ratings aggregators may have an increasing influence over where that money goes. A single film critic can no longer make or break a film, but perhaps thousands of critics, both professional and amateur, can how well can these platforms be trusted especially if they are showing the ratings and making money by selling the tickets? Do they have a bias by rating movies higher than it should be? The objective of the analysis is to provide brief research of one such online movie-ticket selling company, Fandango. The data collected from Fandango are analyzed and compared with the data collected from sites like Rotten Tomatoes, IMDB, and Metacritic. With the help of visualization, fraud and risk can be reduced offered by such online sites.

In the last couple of decades, efforts were made to accomplish a concordance between indoor air quality, air distribution, and energy efficiency and similarly in the center of thermal comfort in the indoor atmosphere. We have shown up at a time when air conditioners add to a significant piece of the power interest in the structure cooling sector. In India, about 70% of the power request is satisfied by thermal power plants, and from this time forward, more energy consumption means more coal consumption causing higher emissions of Earth-wide temperature boost gases. At an outside temperature of 40 °C or above, air forming system plays a basic capacity in keeping up the inside temperature. Global warming has increased the temperature of the atmosphere and that in the end has caused more conspicuous use of air-shaping systems in building cooling. It has become essential equipment to control the internal thermal comfort of business and residential buildings around the globe. The essential concern of any air conditioner is to keep up, with the indoor air quality, temperature consistently, humidity, and air velocity; all these atmospheric properties add to the state of thermal comfort. In this paper, energy saving through semi-indirect evaporative cooling of a vapor compression refrigeration system is studied.

As rapid increase in IoT framework in edge computing, the data storage requirement is also increased. This data is stored in RAID 5 (Redundant array of independent drives) disks which require memory testing and a repair algorithm for a reliable design system. MBIST design system depends on the memory testing algorithm. Various fault detection approaches are introduced using March test on SRAM and DRAM. But, still, some focus is required in terms of time penalty and fault coverage. This paper introduces a novel approach for fault detection in memory with less time penalty, better fault coverage, and device utilization performance with the help of the IoT framework in edge computing. An IoT framework system is proposed for proper monitoring of memory under test for fault occurrence and number of fault repair for SRAM. Results show the optimal faults repair and with less time penalty. The simulation is conducted on MATLAB and Xilinx ISE suite. Proposed work can be used in commercial and space application where radiation hardened memories is used.

This paper investigates the performance of a high electron mobility transistor (HEMT) with a 0.4 µm gate size enhancement mode. The device is composed of an InGaAs/InAlAs structure grown on an InSb substrate, with heavily doped In0:6Ga0:4As source/drain (S/D) regions and dual δ(sigma)-doping linear layers. The transistor described in this paper incorporates a buried Au metal gate technique to minimize short channel effects and enhance transconductance. The device also features heavily doped In0:6Ga0:4As source/drain (S/D) regions, Si dual sigma-doping linear layers at the edges of the In0:75Ga0:25As channel area. The high electron mobility transistor (HEMT) InSb/InGaAs/InAlAs provides outstanding high-frequency performance. Silvaco TCAD simulations that use the accurate methodology at room temperature indicated that the investigated device exhibited good pinch-off performances of IDS = 222.8 A at VGS = − 0.6 V, with a high transconductance of 894.8 A/V and a threshold voltage (IDS) of 3 V.

Diabetic retinopathy (DR) becomes a sight-threatening complication because of diabetes mellitus which affects the retina. Initial identification of DR turns out to be a significant one as it might cause permanent impaired vision in the late stages. The automatic grading of DR seems to have effective benefits in solving such impediments, like rising efficiency, scalability, and coverage of analyzing process, extending applications in developed areas, and enhancing patient prevention by offering premature diagnosis and referral. In recent times, the performances of deep learning (DL) systems in the analysis of DR are close to that of expert-level diagnoses for grading fundus images. This article introduces an Ant Lion Optimizer using ALODTL-DRG technique on retinal fundus images. The presented ALODTL-DRG model performs preprocessing via interpolation image resizing, weighted Gaussian blur, and CLAHE-based contrast enhancement. For feature extraction, Inception with ResNet-v2 model is utilized in this study. At last, the ALO algorithm can be exploited as a hyper parameter tuning strategy to accomplish enhanced DR detection performance. The experimental assessment of the ALODTL-DRG method can be tested by making use of benchmark datasets. A widespread comparison study stated the enhanced performance of the ALODTL-DRG model over recent approaches.

For both military and civilian vehicles, ballistic safety systems frequently use high-strength Titanium and Aluminum plates. The choice of alloy is therefore based on its intended usage, ballistic performance, and safety for them. In this study, the effect of a bullet on Titanium and Aluminum alloy plates with fixed edges on both sides is examined. The effects of bullet thickness and impingement angle on Titanium and Aluminum alloy plates were examined using simulations. These simulations were carried out in Ansys Workbench using the Finite Element Method. Simulations using both material models also revealed a distinct variation in the plate’s deformation. The targeted plate was impacted at a 45° oblique angle with velocity of 830 m/s in every test. The findings demonstrated a crucial slant angle of 45° where the piercing operation transitions to ricochet. The other simulation was run to ascertain the plate thickness where the piercing operation transitions to embedment.

The nonrenewable energy sources give continuous more electrical energy when compared to the renewable energy systems. But the availability nonrenewable energy sources are very less. Also, the nonrenewable energy sources are not safe for the human life. Now, most of the electricity generation companies are working on renewable power supply. The most commonly utilized renewable source is solar. The features of solar are free of cost availability and less effect on human life. But, it gives nonlinear power curves. So, the obtaining of peak power and voltage from the solar system is quite difficult. Here, the Perturb & Observe (P&O) along with Particle Swarm Optimization (P&O-PSO) method is interfaced in the photovoltaic (PV) system for finding the actual working point of the PV module. The proposed topology is studied by using a MATLAB/Simulink window.

The progress in wireless technology has simplified and streamlined the transfer or sharing of data, thereby maximizing its impact on societies worldwide. However, with these advancements, more memory space is needed to store the vast amount of information being transferred. To achieve this, the size of devices must be reduced, necessitating the scaling of MOS transistors to deep submicron levels. Transceiver is being one of the crucial components which is responsible for transmitting or receiving information from the wireless device. Within the wireless transceiver, the frequency synthesizer produces an stable output frequency and further mixed with the received signal down to lower frequencies and vice versa. To operate at high frequencies between 12 and 40 GHz, where operations are carried out at high speeds and coverage is done with multiple beams, circuits must be compatible with high speed. In this paper, we study the VCO component, list the design parameters, and propose a model in which the inductor is replaced with a gyrator-C active inductor to minimize the overall area and use the frequency of oscillations as per the requirement. The design is simulated at 90 nm technology on ADS design tool.

Many real-world classification problems involve changing events where one class has comparatively fewer samples called minority class which is more important to detect. Consequently, the dataset is often unbalanced and shows significantly skewed data. Since the majority class dominates the learning process and tends to sketch all predictions, the conventional classification model leads to biased results where it may easily display excellent performance in the dominant class and bad performance in the minority class. Additionally, the traditional accuracy score is inaccurate since it assigns equal weight to actual positives and actual negatives. This study is aimed to present an empirical analysis of the data imbalance effect on classification algorithms. Six popular and effective data balancing techniques are applied to eight benchmark skewed datasets from the KEEL and Kaggle repositories to analyze and compare the performance.

Miniaturization and multi-band operability have been the key desirable characteristic features of smart antennas designed in wireless communication. Printed folded dipole antennas find their applications in radio frequency identification tags. In this paper, a printed planar folded dipole antenna operating at 2.4 GHz is investigated to meet the requirements of miniaturization and multi-band functionality at 2.6 and 5.6 GHz. Altair Feko is used for the simulation of the antenna. The parametric study is performed to study the change in impedance bandwidth for varying dipole trace width and substrate thickness. The proposed antenna is found to be potential candidate for WLAN applications with dual-band functionality and omnidirectional radiation pattern.

After the COVID-19 pandemic, people began to fear that the monkeypox virus will be the next epidemic. The World Health Organization (WHO) has reported that monkeypox outbreaks have taken place in different regions of Central and West Africa throughout the years, with the latest outbreak being reported in Nigeria in May 2021. Fever, swollen lymph nodes, dry cough, and red rashes all manifest as signs of the monkeypox virus. Most of the symptoms of Measles and chickenpox are comparable. These disorders are given only methodical treatment by the doctor. In Image processing, feature extraction techniques are used to transform raw pixel values of an image into a set of features that can be used for further analysis, such as object recognition, image classification, and image retrieval. Several feature extraction techniques are employed to determine the disease from the images in order to determine which technique works best for this collection of monkeypox skin images (MSID). Wavelets fused with gray-level co-occurrence matrix (GLCM), Haralick features, and local binary pattern are the various feature extraction techniques (LBP) applied in this work. For the classification of various pox virus diseases such as measles, chicken pox, and monkeypox, various classification algorithms such as Random Forest Classification (RF), Naive Bayes (NB), K-Nearest Neighbor algorithms (KNN), Support Vector Machine (SVM), Ada Boosting (AB), and Gradient Boosting (GB) are used in this work. In this paper, four evaluation metrics are used to determine the best feature extraction method for the monkeypox, chickenpox, and measles datasets. Wavelets fused with GLCM produce the highest accuracy (84.41% for gradient boosting and 83.87% for random forest) when extracting features from Monkeypox Skin Image Datasets (MSID).

This study examined how the measurement of eye activity is done using various technologies. Eye tracking (ET) is dominant and prominent. It analyzes the gazing direction of people and measures the entire functioning of the eye. In this paper, we will come to know about various technologies for eye tracking and can able to choose the best approach. Irrespective of technology, as the eye is the most affected part, we can employ this technique in all fields. The main motto of this is to demonstrate a full-fledged review of diverse topics and techniques used in eye tracking. We will be seeing some interesting techniques like skin electrodes, contact lenses, head-mounted and remote systems, and the approach behind them. Most importantly we will learn what is pupil center corneal reflection technique (PCCR). ET gives numerous application which relates to the interaction between humans and computers. This paper also incorporates various elements which took part in the selection of a particular eye-tracking method.

This paper focuses on the development of a lightweight and easy-to-use robotic arm that can be controlled using speech recognition technology. The arm can be controlled using live voice instructions, and the system has been designed to work with both English and Kannada voice commands. The image model used for speech recognition predicts audio labels based on spectrograms, and the system transfers the prediction to the microcontroller using a serial port connection. The microcontroller then moves the arm based on the instructions received. The system is designed to be accessible to people of all backgrounds, regardless of their educational level or other factors. The project has the potential to make a significant impact by enabling people to use robotics to perform tasks that would otherwise be difficult or impossible. The work describes the use of four predefined English and Kannada commands to assess the success rate of speech recognition using machine learning models, specifically the ANN and the CNN. The accuracy of the models was evaluated and compared for both the English and Kannada datasets.

Predictive maintenance (PdM) techniques can increase industrial productivity and reduce maintenance costs by predicting the remaining useful life (RUL) of complicated machines. However, PdM systems involve industrial internet of things (IIoT) devices and machine learning (ML) algorithms, which are prone to adversarial attacks. In this work, first PdM is developed based on four ML classification models: Random Forest (RF), Light Gradient-Boosting Machine (LGBM), Gaussian Naive Bayes (GNB), and Adaptive Boosting (AdaBoost) classifier. Second, the robustness of the ML models under three adversarial attacks is evaluated, using the NASA turbofan engine dataset: Zeroth Order Optimization (ZOO), Universal Adversarial attack, and HopSkipJump attack. Results indicate RF as the most efficient classifier, reaching 96.35% of classification accuracy. Moreover, RF is proven to be the most robust of the examined classifiers under the considered attacks, displaying comparative resilience of up to 83.58% higher, compared to other models.

In the field of industrial environment, prosthetics, rehabilitation, space applications, medical applications, etc., the implementation and use of the robotic hand is of significant prominence to achieve reasonable accuracy and improve production. For perfect grasping, dexterity manipulation and shape adaption the concept of underactuation is an appropriate approach. This paper implements the modelling of a four-fingered robotic hand and the analysis of the material used in the hand during grasping. The proposed robotic hand consists of a four underactuated fingers that are alike to each other. Each finger consists of three phalynx and three joints. Under actuation is carried out by tendon and pulleys. The proposed robotic hand has four fingers with a total of twelve numbers of joints with a total of sixteen degrees of freedom. Solid work platform is used to model the hand and a finite element-based analysis is performed to analyse the various mechanical parameters based on the materials used in the model. The robotic hand holds a cuboid with its fingers. The analysis is accomplished concerning three types namely deformation, stress and strain test. The analysis provides significant data about the parts of the robotic hand that can be destroyed if subjected to greater force. Therefore, this could be useful for a major design upgrading and confirming the suitability of the robotic hand in actual application.

Automation is currently necessary in many industries including the core. After studying issues in various industries, we discovered that the majority of errors take place in tasks that require human intervention; in fact, the product’s quality and turnaround time must be maintained. By automating the industry completely or partially, these can be maintained. Our main aim is to create a six-axis robotic arm that can fully or partially automate industry and provide a production line with a variety of benefits. They have better wrist movements and are more elastic, which helps to improve movement and the production line. We utilized an Arduino Nano for automatic control and an HC05 Bluetooth module for manual operation to complete this project.

Robots are widely used to carry out various tasks in different industries worldwide. The movement of a robot is necessary for any task accomplishment. While moving, a robot must prevent collisions with obstacles to reach its destination successfully. The motion control algorithm governs a robot’s movement. One such method is Lyapunov-based control scheme (LbCS). LbCS is a popular method for controlling a robot’s motion, but the technique suffers from a problem known as collinear. This problem occurs when a robot, an obstacle, and a target are in a linear position, which gets the method trapped into local minima. This paper tackles this problem using a heuristic-based method, ant colony optimization (ACO). The ACO will be activated when the LbCS gets trapped in local minima. This paper presents an algorithm, ACO-LbCS, that solves the collinear problem of LbCS. This hybrid algorithm has been strategically formulated using ACO and LbCS. The algorithm has been applied to multiple obstacle’s environment. The results show that the problem of local minima has been solved by the proposed algorithm.

Due to rapid enhancement in image processing, there is need to design and implement an improved edge detection algorithm in order to analyzing the edges of an original image. Optimization mechanism based on ant colony optimization technique has been used in present work. Research work is focused on implementation of edge detection using ant colony optimization algorithm on MATLAB and to improve the drawbacks of that algorithm and comparing it with the new improved algorithm. Present research is focused on the performance parameters, namely RMSE and PSNR. Thus, edge detection process The edge detection process considers selection of the image as input, and image is saved in a 256 color bitmap format. Then edge pixel values and generated the edges in image is calculated to generate the results with improved quality edges. Finally, comparison of the results of both algorithms and represent those results are made graphically. Proposed research is supposed to play significant role in area of image processing and quality enhancement.

The COVID-19 pandemic has caused significant fear, anxiety, and complex emotions or feelings in a large number of people. A global vaccination campaign to end the SARS-CoV-2 epidemic is now in progress. People’s feelings have become more complex and varied since the introduction of vaccinations against coronavirus. The use of social media platforms such as Twitter enables users to communicate with one another and share information and perspectives on a wide variety of topics, spanning from local to international concerns, from global to personal. Twitter will prove to be a helpful source of information that can be tracked regarding views and sentiments regarding the SARS-CoV-2 vaccination. To better understand public views, concerns, and emotions that may influence the achievement of herd immunity targets and limit the pandemic’s impact, this study uses deep learning to identify the themes and sentiments in the public about COVID-19 immunization on Twitter. Moreover, this paper consists of a detailed explanation of the sentiment analysis with their challenges, classification, approaches, applications, and VADER.

The fundamental goal of sentiment analysis is to find and categorize any views or feelings that are communicated in a text. Nowadays, discussing thoughts and expressing feelings through social networking sites is widespread. Consequently, a vast amount of data is generated every day, which can be mined successfully to extract valuable information. Performing sentiment analysis on such data can be useful for producing an aggregated view of particular products. Due to the prevalence of slang and misspellings, sentiment analysis on Twitter is frequently a challenging undertaking. Additionally, we are constantly exposed to new terms, which makes it more difficult to assess and compute the sentiment compared to traditional sentiment analysis. Twitter limits a tweet's length to 140 characters. Consequently, obtaining important information from brief messages is another obstacle. Knowledge-based approaches and machine learning can significantly contribute to the sentiment analysis of tweets. The amount of data produced by people, i.e., users of a certain social site, is growing exponentially as a result of changing behavior of various types of networking sites like Snapchat, Instagram, Twitter, etc. The purpose of this paper is to determine the emotions underlying these posts. We have decided to use Twitter as our platform for this. In this study, we investigate the views expressed by Twitter users concerning certain companies. By computing a basic sentiment score and then categorizing them as positive or negative, the corporation would be provided with critical feedback about its products from individuals around the world. The proposed LSTM model has proved to be 93% efficient in comparison with previous models which were accurate up to 86%.

Alzheimer’s disease is the most common forms of dementia. Dementia is the general term for cognitive decline severe enough to impede with daily activities. Early diagnosis of Alzheimer’s disease is important for slowing or stopping the disease’s development, and experts can start preventive treatment right away. The experts must be capable of identifying Alzheimer’s disease in its earliest and most challenging phases. The fundamental objective of this study is to create a machine learning model that can automatically diagnose disease using MRI, a widely used diagnostic tool. This research employed structural MRI to find out the difference between patients with Alzheimer’s disease (AD), stable mild cognitive impairment (sMCI), progressing mild cognitive impairment (pMCI), and normal cognitive functioning (CN). In this research paper, machine learning models, namely SVM, RF, DT, and CNN, are used for multi-class classification. CNN obtained the highest testing accuracy of 88.84% among the four models, with a precision of 80.42%, a recall of 73.17%, and an F1-score of 76.62% for the CN versus sMCI versus pMCI versus AD multi-class classification.

Estimating household power consumption energy usage patterns can assist households in planning and managing their power consumption. To address elaborate time-series data, long short-term memory artificial neural networks are a promising strategy. However, a decomposition-forecasting method called variation mode decomposition is necessary to handle challenging time series. The accuracy and effectiveness of machine learning models are influenced by their hyperparameter values. This paper suggests using a altered sine cosine algorithm to optimize the hyperparameters of the long short-term memory model. This algorithm enhances the accuracy and performance of household energy consumption forecasting. The proposed model is compared to other long short-term memory models that are optimized by advanced metaheuristics. Simulation results indicated the improved sine cosine algorithm surpassed other advanced approaches in terms of standard time-series forecasting metrics.

A four-element planner, dual-band MIMO antenna for 5 generation (5G) smartphone application was proposed which can be implemented in wireless handsets. In this research, the design of an H-shaped monopole antenna was developed and simulated using computer simulation technology (CST) software. The antenna was fabricated on an FR4 substrate with the dimensions of 150 × 75 × 0.8 mm3. The two 5G new radio bands, n79 band (4.4–5 GHz), and LTE band 46 (5.1–5.9 GHz) are covered by the antenna without using any additional decoupling structure. The performance characteristics of the antenna were analyzed, including the reflection coefficient, radiation pattern, envelope correlation coefficient, and efficiency. The antenna displayed excellent characteristics, such as good impedance matching (return loss > 10 dB), high isolation (> 18.8 dB), high efficiency (> 60%), and low envelope correlation coefficient (ECC, < 0.03) across the operating frequencies. The proposed MIMO antenna is simple and compact, leaving enough space inside handheld mobile terminals for the integration of other circuits.

It is very difficult to obtain the optimal parameter of controllers for load frequency control (LFC) problem of a multiarea power system in deregulated environment. Deregulated power system contains multisources and multistakeholders; therefore, conventional LFC methods are not effective and competent. The primary goal of LFC in a deregulated system is to restore the frequency to its original value as soon as feasible while also minimizing uncontracted power flow in tie line between neighboring control regions and tracking load balancing contracts. Gains of PID controller are required to be optimized in order to fulfill the objectives of LFC. Here genetic algorithm as well as particle swarm optimization techniques are presented in this paper for optimization of controller parameters in order to achieve the purposes of LFC of two-area deregulated system taking suitable objective function that are to minimize the frequency deviations of both the areas and to maintain tie line power flow according to contractual conditions. System has been simulated under MATLAB/Simulink, and dynamic responses have been obtained for many contractual conditions between GENCOS and DISCOS. It is confirmed by the results that the soft computing–based PID controllers are capable of maintaining the frequency in the pre-specified range and keep the tie line power flow as per the contractual conditions. An analysis has been done by comparing the dynamic responses of the system with PSO-based controller and GA-based controller.

Over the last few years, we have witnessed a relentless improvement in the field of computer vision and deep neural networks. In a deep neural network, convolution operation is the load bearer as it performs feature extraction and dimensionality reduction on a large scale. As the models continue to go deeper and bulkier for better efficiency and accuracy there is a rapid increment in storage requirements too. The problem arises when performing computation with efficient numerical representations for embedded devices. Transitioning from floating-point representation to fixed-point could potentially reduce computation time, storage requirements, and latency with some accuracy loss. In this paper, an analysis of the effects of quantization of the first convolutional layer on the accuracy, and memory storage requirement with varying bit-width for fixed-point integer values of network parameters has been carried out. The approach adopted is post-training quantization with a mixed-precision format to avoid model re-training and minimize accuracy loss by using root-mean-square-error (RMSE) as a performance metric. Various combination has been analyzed and compared to find the optimal precision to implement on a resource-constraint device. Based on the analysis, the suggested bit-width of I/O data for this implementation is selected as <10,5> and mid-data be <20,10> instead of <16,8> and <32,16> respectively. This combination of bit-widths has reduced memory consumption such as BRAM by 10%, DSPs by 98.6% and FFs by 40.27% with some accuracy loss.

The purpose of this research is to employ non-integer order calculus to enhance the control action of the non-linear fractional order fuzzy PD plus I (FOFPD + I) controller. To operate a non-linear 6-DOF Puma 560 robotic arm, a FOFPD + I controller is developed and implemented. Fractional order fuzzy proportional derivative (FOFPD) and fractional order fuzzy integral (FOFI) controllers are used to create the proposed controller. Because of the non-linear gains, the proposed control approach preserves the linear structure of the fractional order proportional derivative plus integral (FOPD + I) controller while still providing adaptive capabilities. Further, the PID controller is also derived to compare with the FOFPD + I controller. Both, FOFPD + I and PID controller’s parameters are optimized using non-dominated sorting genetic algorithm-II (NSGA-II). The performance and effectiveness of the presented controller are examined in terms of trajectory tracking, tracking error, and robotic arm control efforts. Simulation shows excellent performance of FOFPD + I controller over traditional PID and fuzzy logic controllers. To be precise, on the current optimal solution, suggested controlling strategy is more than 80% efficient than other strategies.

The capacitated vehicle routing problem (CVRP) is a significant topic in distribution networks. CVRP is essentially a subset of the vehicle routing problem (VRP). As CVRP has many applications in transport, logistics, and telecommunications, exact methods are not properly suitable for finding the optimal solution to large-scale CVRP problems, so most of the researchers are focusing on meta-heuristics like genetic algorithms and ant colony algorithms. As the CVRP is an NP-hard problem, it means that an efficient algorithm for solving the problem optimally is unavailable. In this paper, approximate optimal solutions to the CVRP are generated using a two-phase method with the modified genetic algorithm. Phase one includes clustering, and the second phase is based on the genetic algorithm, which is a meta-heuristic algorithm used for finding the best solution. The proposed method is tested on a set of benchmark instances from the literature. We report computational results with this meta-heuristic algorithm on some instances taken from the literature.

In recent developments, a lot has been done for computer vision applied to human action recognition and violence detection. Although various datasets are available for action and violence recognition, there is a clear lack of datasets that include non-violent and violent activities simultaneously from an aerial view. A new aerial video dataset for concurrent human action recognition, including violence detection, is presented in this study. It consists of 60 min of fully annotated data with two action classes, namely violent and normal (non-violent). The current dataset addresses various factors that are not considered in the existing datasets, like changes in the altitude of the drone, changes in the angle at which the video is being captured, video captured during motion, changes in frame rates, videos from different cameras with different configurations, multiple labels for every subject, and labels for violent activities. The resulting dataset is a multifaceted representation of the real-world scenarios, which addresses various shortfalls in the existing datasets. The current dataset will push forward computer vision applications for action recognition, particularly automated violence detection in real-time video streams from an aerial view. Furthermore, the curated dataset is validated for violence detection using machine and deep learning algorithms, namely Support Vector Machine (SVM), Long Short-Term Memory (LSTM), Bi-Directional LSTM (Bi-LSTM) and Adaptive Boosting (AdaBoost).

This research work is on designing an optimum and robust controller to improve the infusion of drug in patients during post-operative conditions in order to automatically control the mean arterial pressure. The control of mean arterial pressure of critically-ill patients is a prominent research field due to its ability to lessen recovery time, lower medical costs, and better medical staff management by lowering their burden. However, due to uncertainty in patient’s sensitivity toward the drug infused and external disturbance/noise, this is a complicated task. In this work, two efficient controller schemes, interval type-2-fuzzy logic controller-proportional integral derivative and fractional order-interval type-2-fuzzy logic controller-proportional derivative + proportional integral controller are presented. For effective optimization of these control schemes, one of the most effective nature inspired optimization, cuckoo search algorithm is implemented for finding their optimal values of parameters. In the work, simulation results and performance index i.e., settling time, overshoot IAE values shows superior performance of fractional order-interval type-2-fuzzy logic controller-proportional derivative + proportional integral controller over IT2-FLC-PID controller.

Alzheimer’s disease (AD) is a slow-paced irreversible brain disease and a neurodegenerative disorder that accounts for approximately 70% of the dementia cases estimated worldwide, the number of which totals to more than 46 million. AD affects the brain's thinking capacities along with significant memory loss. People with onset of aging are found to be more prone, with greater memory loss, cognitive difficulties, etc. Currently, there exists no fixed cure for AD, but early detection and characterization are proven to be helpful. Methodologies like electroencephalograms (EEG), magnetic resonance imaging (MRI), computed tomography, positron emission tomography (PET) scan, etc., are helpful in providing information regarding the persisting conditions of the brain cells. Computer-aided diagnosis (CAD) along with biomedical data processing when applied to machine learning and deep learning methodologies has vastly helped sophisticated techniques like CNNs, SVMs, etc., to evolve and achieve promising prediction accuracies. This paper provides a review and critical evaluation of recent research on early detection of AD using ML techniques which employ a variety of complex optimization and statistical techniques to obtain a better accuracy score. Along with advancement in computational capabilities, other factors such as preprocessing and feature extraction along with class imbalance have distinctively helped improve the prediction score which has overall helped produce better prediction with respect to earlier detection of AD.

The subject of this research is navigation of a compartmentalized robot made up of several structures which are car-like with a global rigid formation. The system must move safely to a predetermined target in an already-known workspace that is filled with obstacles. A target convergence and obstacle avoidance method is put forth that is effective for any quantity of obstacles. The smallest distance between an obstacle and the point closest to it on each line segment that forms the rectangular protected zone of the car-like units can be calculated analytically using the minimum distance technique, as proposed in this paper. The Lyapunov-based control scheme (LbCS) is used to construct continuous acceleration-based controls. The computer simulations are presented to validate the proposed controllers, and the system demonstrates the rigorous upkeep of a rigid formation. The successful presentation of controllers opens further research in developing and applying controllers by considering various compartmentalized robots in swarming models, splitting and rejoining units, and applying rotational leadership concepts.

This paper presents a design of velocity controllers for a 2-link dual-arm mobile manipulator which is required to move from its starting configuration to a final position while targeting to avoid multiple fixed circular obstacles of random sizes and positions, and observing all mechanical singularities which are associated with the system. With the help of Lyapunov-based control scheme (LbCS), nonlinear time-invariant continuous velocity-based control laws are formulated which enable the center of the car-like mobile structure to converge to a predetermined target position and the links attain a final orientation. The method also guarantees stability associated with the system proving the use of direct method of Lyapunov. The computer simulations illustrate the effectiveness of the proposed technique.

There can be serious and harmful effects on people following a fall event due to its severity. This paper presents a real-time fall detection using sensor fusion to improve the overall accuracy of the system when subjected to continuous operation. The system combines data from accelerometer and an ultrasonic sensor to detect falls in real time. The developed mobile application accurately identifies fall scenarios and sends SMS notifications to emergency contacts. The proposed system's ultrasonic sensor module has wireless communication capabilities, while the accelerometer readings are acquired from the smartphone. Appropriate feature, threshold values, and program flow have been chosen, such that fall detection is accurate, and the system is operational even if one of the sensor malfunctions. The proposed system has been validated experimentally and the accuracy, sensitivity, and specificity are 87.5%, 89.47%, and 94.74%, respectively.

Gait recognition has become more popular and significant in the recent years due to security concerns since it can be carried out remotely without authorization. This article discusses the vision-based model and model-free feature extraction methods for identifying human gaits. Both methods are distinctive in and of themselves. The structural parts of the human body are dealt with via model-based approaches, including joint locations, joint angles, stride length/cadence, and 2D stick figures. Model-free techniques, including gait energy image, absolute frame difference image, gait history image, etc., give spatiotemporal information on gait silhouettes. Subject identification is made based on the high rate of recognition after approach-wise features are provided to classifiers. The important characteristics will then stand out.

Cardiac disease diagnosis is very important domain in biomedical and technological innovations. The right ventricle (RV) volumetric analysis is useful for finding anatomical and functional defects and blood loading capacity of heart at right side. The volumetric measurements of RV such as blood volume at systole (ESV) and diastole phases (EDV) are significant for further decisions in cardiac disease diagnosis. For experimentation and to find better one, we have been performed three techniques A. motion-based clustering B. intensity-based clustering C. deep learning-based architecture. These approaches are used for measuring volumetric parameters such as end systole (ESV), end diastole (EDV) and ejection fraction (EF). Then, results are compared with ground truths provided by clinicians. The minimum percentage error occurred in measurements of deep learning techniques as compare to others. The clustering approaches having limitations of data scarcity, which can be eliminated using DL techniques and shows better performance.

Exploitation of massive amount of multidimensional data of numerous diversities from heterogeneous sources is emerging. Segments can be expanded by the effective use of the data and simultaneously the warehoused data furnishes significant eventualities for strengthening cardinal decision making. Cutting edge intelligence and efficacious approaches and methods are essential to produce a model by means of various types of data. Machine learning is described as constructing a model for handling unfathomable accumulated or streaming data which is difficult to be handled by conventional data processing techniques. The prerequisite for machine learning is also prompted based on preprocessing, analysis, demonstration and prediction on diverse categories of data. The experimentation has been performed on Mushrooms dataset and Census dataset by carrying out machine learning algorithms. It has been proved that the ensemble algorithm shows better performance on the intended datasets in terms of various performance measures.

Predictive modeling is a new area of data science and machine learning that is gaining popularity. It provides sustained business growth, accurate future predictions, and trend estimations. Predictive modeling is the process of creating, processing, and validating a model that may be used to make future predictions using known results. Predictive modeling depends on the complete and precise datasets, however some of the datasets are imbalanced in nature that leads to data misclassification. Models trained on an imbalanced dataset with a small number of minority class instances, despite their high accuracy, would perform poorly during training. In this paper, an approach for synthetic minority class data generation using agglomerative hierarchical clustering and ward’s linkage criteria is proposed. Experimentation is carried out using five real-world datasets, namely Abalone, Page Blocks, Pima, Vehicle, and Yeast available at KEEL Data Repository. Testing of the experimentation is done using the SVM classifier with radial-bias kernel function. Data visualization is performed for understanding statistical properties, correlations, and distribution of class instances in the feature space. The classifier model is evaluated for before and after synthetic data generation using f-measure, recall, precision, and accuracy.

Mankind cannot thrive without automobiles due to the greater productivity they provide. Traditionally, most of the energy used to power automobiles comes from fossil fuels. These vehicles are being replaced by electric vehicles over time. Typically, a car's braking system uses hydraulic braking technology. However, because it generates more heat when braking, this conventional braking technique wastes a lot of energy. Therefore, the development of regenerative braking introduced in electric vehicles has eliminated these drawbacks, in addition to assisting in energy conservation and increasing the vehicle's efficiency. When operating in regenerative mode, the motor transforms kinetic energy into electrical energy to recharge the batteries or capacitors. The hill climb algorithm is used to analyze the regenerative braking of EVs, and MATLAB software is used to create and model electrical circuits for the required configuration for regenerative braking. The parameters of lead acid battery as well as parameters of DC machine have been analyzed. The performance of the regenerative braking model with hill climb algorithm is analyzed against the regenerative braking without any algorithm which is considered as the basic model.

Facebook, Twitter, and Instagram are just a few examples of how social media have changed our lives. People are more linked than ever before, leading to the development of a distinct online identity. Recent studies have revealed that an increased number of hours spent on social media platforms is connected with an increased likelihood of developing depression. Depression is characterized by pervasive sadness and a general absence of interest in most activities. Severe depression, often known as major depressive disorder, is a serious mental illness that can have far-reaching effects. The purpose of this study is to analyze depression, utilizing a variety of socio-demographic and psychological data to determine if a person is depressed or not. Different operations have been performed, including data collection, preprocessing, feature selection, classification, and evaluation. This research is evaluated on the depression detection dataset. Data is processed in the data preprocessing step by checking null and missing values and performing data encoding using a label encoder. Further, the recursive feature elimination technique has extracted the most important features from the dataset in the feature selection. On the other hand, machine learning-based SVM and DT techniques are used for classification. The performance of these models is measured using different performance metrics. After applying these methods, the proposed decision tree model obtains the highest 98% accuracy, which is better than the other models.

Many problems related to instability are faced by power system which crates huge oscillations and make a system unstable. Such problems can easily overcome by a good damping controller. To find out the solutions of stability problems many methods including analytical methods or the numerical-based have been applied in many ways to obtain extreme values. And now these methods are developed into the more advanced form of themselves known as optimization techniques. To solve complex engineering design problems and real application, GWO is one of the more suitable optimization techniques. Applications of GWO algorithm are investigated in this paper to tune the parameters of proportional integral derivative-structured TCSC-based controller to damp out the power system oscillations and improves settling time subjected to various loading conditions. Dynamic performance of proposed controller is analyzed for SMIB power system using MATLAB/Simulink. An obtained simulation result shows the performance of GWO-tuned PID-structured Thyristor Controlled Series Capacitor-based controller and compare with previously published non-dominated sorting genetic algorithm-II (NSGA-II) and cuckoo search algorithm (CSA) for the same power system.

In the present work, an adaptive neural controller is designed and applied to pressure control in industrial processes, implementing artificial neural networks and adding an algorithm based on adaptive interaction theory to them, with which it is possible to obtain an intelligent controller. The controller based on neural networks has adaptive properties through the new Brandt-Lin algorithm; this will allow controlling the process without a training phase and prior knowledge of the plant. Therefore, the intelligent controller can adapt online to changes in industrial processes. The challenge of this work is the implementation of this intelligent controller in a real plant (Festo MPS PA Compact Workstation), whose study will be carried out in a training kit that will simulate the industrial process to be controlled. Finally, the work will demonstrate the supremacy of the proposed controller compared with a classic PID controller, being far superior in all the simulations carried out in the control of the real plant.

This project proposes a real-time robot that can interact with humans based on the gestures fed to it as input. The proposed proposal aims to develop a constructive design of a robot that has computer vision and is trained to read human gestures. There is a need for intelligent robots in the healthcare industry. The impact of this project will be on sophisticated healthcare systems, especially the surgical system. The implementation is achieved by training the robot using deep CNN and making the robot perform certain functions like moving the arm upwards and downwards as well as opening and closing the robot grippers. It is also important to mention that a comparative analysis has been made with the existing system and advanced technology called MediaPipe framework for the acquisition of input signals. The comparative analysis will give us a clear picture of the usage of different types of classifiers for training robotic models. With the impact of this project, it would be easy for the physicians to pick and place the medical equipment in a correct manner and provide assistance to the surgeon during surgery. This device can also be very useful in robot-assisted surgeries as it can be further developed to perform actions like drilling and making incisions.

In the scientific community, there is a lot of interest in studying Mars with tiny meanderers that can travel great distances and carry a few scientific instruments. In order to find instruments against outcrops or free shakes, scan a region for an example of interest, and gather rocks and soil tests for return to Earth, such meanderers would travel to locations that were separated by a few kilometers. Within the mission’s constraints of mass, power, volume, and cost, our research objective is to develop innovations that make such situations possible. For data on the planet’s climatic history, fixed-landers will provide excellent, logical information about the air and dirt. As we are executing the damper suspension to diminish the vibrations brought about by the meanderer when it is moving or moving all over the world, the wanderer can convey payload more than 10kg upon its back, we carried out the pick and spot arm to pick the examples for the lab research, we executed the rocker-bogie system as the wanderer can move all over the planet in any landscape.

In the recent past, swarm robotics technology has been widely applied in the variegated industrial domains. It essentially incorporates the multi-robot system with robots communicating with each other. In this work, a swarm of 30 robots is considered for warehouse management applications. A particle swarm optimization (PSO)-based swarm robotics system is manually designed and then simulated. The goal of the swarm is to employ the particle swarm optimization technique to efficiently complete loading and unloading duties in a warehouse. The swarm of robots that communicate with one another via radio frequency (RF) communication is subjected to particle swarm optimization. The physical prototype is built with two robot system equipped with sensors such as infra-red (IR), ultrasonic; motor drivers and RF communication unit. The major purpose of this project is to replace conventional approaches for finding the shortest path, such as the A* algorithm and Djikstra algorithm, with particle swarm optimization in order to load and unload items quickly and without the involvement of people. The physical prototype with sensors and RF communication demonstrates the feasibility of the proposed approach and provides a basis for further experimentation and improvement. Further research and experimentation are necessary to address the challenges and limitations of swarm robotics in warehouse management.

Additive manufacturing is a recent trend in production processes owing to its sustainable approach. As a process in itself, additive manufacturing represents a more sustainable means of production as it eliminates the use of excess material and thus unnecessary waste. Wire and ARC additive manufacturing (WAAM) process is an important metal additive manufacturing process. The improvements in surface quality and dimensional accuracy are critical for WAAM. In the current work, two contemporary artificial intelligence (AI)-based algorithms, teaching learning–based optimization (TLBO) and particle swarm optimization (PSO), are applied for optimizing the five process parameters such as wire feeder, pulse voltage, frequency, pulse time, and welding speed with four objectives such as current, voltage, heat input, and width-to-reinforcement ratio which are considered as referred from Youheng et al. (Int. J. Adv. Manuf. Technol. 91:301–313, 2017). The results obtained with TLBO and PSO are comparable and improved by 24%, 32%, and 42% for the minimization of voltage, heat input, and maximization of width-to-reinforcement ratio, respectively. Both algorithms are observed to be robust. The convergence analysis for the algorithms is also discussed.

Body sensor network (BSN) is a new technology. BSN care system begins by placing tiny, lightweight sensors on the patient’s body that communicate with one another and the body-connected co-ordination node. This system is primarily concerned with measuring and estimating critical parameters such as ECG, temperature, and blood level. This real-time system focuses on a number of parameters, including patient health, motion detection and data transmission, and message transmission to the primary responder and hospital server. We use four types of sensors in this system: temperature sensor, pulse sensor, oxygen sensor, and fall detection sensor, which collect patient information and send it to the microcontroller. From there, the information is transferred to an android smartphone and server via the internet. We propose an IoT-based health system based on body sensors that meet the requirements effectively.