International Journal of Advanced Mechatronic Systems (10 papers in press)

Regular Issues

• InsuDet: a lightweight insulator defect detection algorithm based on YOLOv8  
  by Jinqing Shen, Hong Ye, Chunjun Tang, Yifu Chen, Yan He, Enhui Zheng 
  Abstract: Failures caused by insulator defects may seriously threaten the normal operation of power systems. Hence, how to detect and maintain insulator defects in a timely and accurate manner is of vital importance. However, the small size of insulator shells coupled with complex backgrounds make such detection an arduous task. To address the issue of low accuracy in detecting insulator defects, this paper proposes InsuDet, an improved lightweight object detection network based on YOLOv8. In order to further reduce the complexity of the model, MobileNetv3 blocks are chosen as the backbone of the model in this paper. Additionally, the bottleneck structure in C2f is replaced with deformable convolution structure and small object detection layers are introduced to enhance the feature-extraction performance of the model, improving its ability to identify dense small objects. Our experimental results show that compared to the YOLOv8-s baseline model, our model has increased precision and recall rates by 2.5% and 1.8%, respectively, reflecting at 96.8% and 95.5%, correspondingly. Furthermore, our proposed model reduces the number of parameters by 21%, and achieves a detection speed of 111.6 frames/second. Our model can accurately and timely detect insulator defects, while its lightweight structure ensures that it can be effortlessly deployed on mobile devices like unmanned aerial vehicles.
  Keywords: YOLOv8 network; insulator defect detection; aerial inspection images; image processing.
  
  
• Artificial neural network-based fault detection and identification in a seven-level fault-tolerant inverter  
  by Aquib Mehdi Naqvi, Pushkar Tripathi, S.P. Singh 
  Abstract: The advantages of multilevel inverters (MLIs) have led to their widespread use in various applications. However, the numerous semiconductor switches in MLIs pose reliability challenges. Researchers have developed fault-tolerant (FT) topologies and operational methods to ensure uninterrupted power supply. Detecting and identifying faults is essential for corrective measures and FT operation. This paper presents an artificial neural network (ANN)-based fault detection and identification method for a seven-level inverter. The MLI possesses high redundancy in switching states, which helps to maintain output voltage during open circuit switch faults. A five-layer ANN is trained with data of total harmonic distortion (THD) and average output voltage under different fault conditions and loads. Identifying the faulty switch allows for adjustments to the switching sequence and angles, ensuring operational continuity and optimal voltage levels. Validation of the fault detection and identification method is confirmed through simulation results.
  Keywords: multilevel inverter; MLI; total harmonic distortion; THD; fault-tolerant; voltage level; neural network; switches.
  
  
• Artificial pancreas design for BG regulation in TIDM patient based on sliding mode controller with sliding hyperplane  
  by Girija Sankr Panigrahi, Akshaya Kumar Patra, Alok Kumar Mishra, Samjeeb Kumar Kar 
  Abstract: Regarding realisation of an AP, specifically for the optimal management of concentration of blood glucose (BG) with an appropriate feedback mechanism has great significance for Type-I diabetes mellitus (TIDM) patients since past few decades. The AP with an efficient control approach algorithm for regulating various continuing clinical illnesses which are dependent on prolonged drugs, is now extremely essential besides the BG control. In the current effort, a brand-new SMC/SHP has been proposed to address the aforementioned issue. A patient with TIDM and AP in a 9th-order state space model is taken into consideration for constructing the SMC/SHP. To enhance control performance, the conventional sliding mode control (CSMC) is reformulated in this control method using a sliding hyperplane parameter basing on the whole state feedback technique. By comparing the results with other previously published control methodologies, comparative result analysis serves as evidence for the justification of SMC/SHP’s improved control performance. The simulations are run in the environment of MATLAB/SIMULINK and the results indicate that the suggested approach has relatively higher ability to manage concentration of BG within the normo-glycaemic range in terms of robustness, accuracy, rapid damping and stability.
  Keywords: insulin; pancreas; glucose; metabolism; sliding mode control.
  
  
• Adaptive inverse control for an optimal pitch-angle controlled horizontal-axis wind turbine test rig  
  by Akrm Abaza, Ahmed Farouk AbdelGawad, Mahmoud Ahmed 
  Abstract: One of the most relevant issues today is the efficient utilisation of renewable energy resources. For the purpose of research and development of control algorithms, a wind turbine test rig with a varying blade pitch-angle is designed, modelled, and implemented. For obtaining the highest power coefficient at various wind speeds, the maximum power point can be tracked by changing the pitch angle. For this purpose, an adaptive inverse controller is introduced. This controller has the advantage that it can update its parameters without referring to a reference model. The proposed non-model based adaptive (NMAC) algorithm is based on inverting an adaptive filter which is placed before the plant to be controlled. Simulation results showed a significant improvement of the proposed controller compared to other traditional PID controllers. Experimental results are carried out using a wind tunnel and show the significant effect of changing the pitch angle at various wind speeds.
  Keywords: adaptive inverse control; blade pitch angle control; horizontal axis wind turbine; renewable energy; non-model based adaptive controller; NMAC; mechatronics.
  
  

Special Issue on: Intelligent Mechatronic Systems and Robotics in Sustainable Development

• Electrostatic discharge test automation using Staubli TX2-90 robot  
  by S. Abhishek, Gokul Jagadish, Shibu Krishnan, Amal Prakash, Arjun R. Nair 
  Abstract: Electrostatic discharge (ESD) testing is a type of EMC immunity test used to make sure that products like actuators, sensors and controllers will continue to function normally in the presence of an electrostatic events rapid release of energy. Growing miniaturisation requires comprehensive product standards. Multi-pin connectors with close arrangements demand meticulous ESD testing. Each pin undergoes multiple discharges, polarities, voltage levels, and RC networks, making manual testing error-prone and time-consuming. Automatic testing offers precise execution, repeatability, and efficiency. Using the Staubli TX2-90 robot with 6 DOF and an ESD simulator gun as the end effector, we developed VAL3 program with SRS software. This allows us to control the robot and simulator. An HMI screen on the teach pendant inputs test parameters and initiates the test, streamlining the entire process. Automatic ESD testing not only ensures quality but also saves significant time and effort in compliance testing for electronic devices.
  Keywords: electrostatic discharge; ESD; Staubli; VAL3; DUT; Staubli robotics suite; SRS; human machine interface; HMI.
  
  
• Enhancing recycling efficiency: a SCARA robot and CNN model for waste segregation  
  by Gokul Jagadish, S. Abhishek, Amal Prakash, Arjun R. Nair 
  Abstract: The development of waste sorting robots, particularly the SCARA robot, has shown promising results in automating the recycling process. By utilising convolutional neural networks (CNNs) for image classification, the SCARA robot can accurately differentiate between recyclable and non-recyclable materials. This technology has the potential to significantly improve waste management practices, enhancing efficiency and accuracy in waste segregation. Future research could focus on integrating additional sensors and implementing advanced algorithms to further optimise the sorting process. Additionally, exploring the potential of deep learning techniques to classify a broader range of refuse materials could further enhance the performance of waste sorting systems. The SCARA robot’s design, which includes a four-DOF base and the use of 3D printing technology, has been carefully analysed and chosen for optimum output and manufacturing simplicity, contributing to its reliability and durability.
  Keywords: SCARA; waste-sorting; CNN; recycling; deep learning.
  
  
• Development of a low cost structured light 3D reconstruction system  
  by G. Karthikeyan, G. Senthilnathan 
  Abstract: 3D reconstruction using structured light scanning has emerged as an indispensable component within the workflows of numerous engineering fields. Structured light scanning has revolutionised design, analysis, and fabrication processes with its ability to capture detailed surface information and produce high-resolution three-dimensional models. By projecting a known pattern of light onto a subject and analysing the distortions caused by its interaction with the surface, this technique enables the precise reconstruction of complex geometries, allowing engineers to visualise, measure, and manipulate objects with unparalleled accuracy. This paper discusses the development of a low-cost and robust structured light scanning system and evaluates its performance based on vertex quality histogram using Hausdorff distance. It also provides an overview of the basic principles of structured light scanning, emphasising the procedure to achieve good calibration results. Its transformative impact on industries such as manufacturing, architecture, medical imaging, and robotics has enhanced the overall process involved. As structured light scanning continues to evolve and adapt to the ever-changing demands of modern engineering, it is poised to further shape the future of design, analysis, and innovation in various industries.
  Keywords: structured light scanning; 3D reconstruction; projector-camera calibration.
  
  
• Smart in-pipe monitoring robot for crack detection  
  by Amal Alex, Pramod Sreedharan, Jeetu S. Babu 
  Abstract: The purpose of a pipeline is to transfer fluids (water, oil, gas or petroleum products) from one place to another place. This smart robot can travel inside the pipe to identify cracks and other potential damages. It is a four wheeled robot equipped with infrared (IR) sensor arrays to identify the cracks and an ultrasonic sensor to detect obstacle while moving through the pipe. The camera provides live video streaming of the pipe. Wheels are tilted at an angle of 45 degree for better stability and friction inside the pipe. It is equipped with Bluetooth module and Wi-Fi connectivity. LED light strips are provided for illumination inside the pipe. It is controlled by using a smartphone or laptop. Two Arduino microcontrollers are used for entire control system. The robot is powered by Li-ion rechargeable batteries. Design and motion simulation are done in SolidWorks 2017. For testing purpose, a 40 cm round aluminium pipe had taken.
  Keywords: smart robot; IR; ultrasonic; cracks; pipe; Bluetooth; camera; Wi-Fi; Arduino.
  
  
• Demonstration and analysing the performance of image caption generator: efforts for visually impaired candidates for Smart Cities 5.0  
  by Rohit Rastogi, Vineet Rawat, Sidhant Kaushal 
  Abstract: Image caption generation has become a prominent area of research due to its potential applications in multimedia understanding and accessibility. This paper presents a comprehensive study of three state-of-the-art approaches for image caption generation, employing convolutional neural networks (CNN) with long short-term memory (LSTM) networks, attention mechanisms, and transformers. The first approach utilises a CNN-LSTM architecture, where the CNN acts as an encoder to extract meaningful visual features from input images. These features are then fed into an LSTM-based decoder, enabling the generation of descriptive captions. The second approach introduces the use of attention mechanisms, allowing the model to focus on specific regions of the image while generating captions. This technique improves the caption quality and ensures that the generated text corresponds more accurately to the content in the image. Lastly, the third approach incorporates the powerful transformer architecture to capture long-range dependencies in the generated captions, enabling better contextual understanding and coherence.
  Keywords: CNN; LTSM; transformer; image caption; attention mechanism; benchmark; NLP.
  
  

Special Issue on: RAISE2023 Mechatronics, Automation and Cyber-Physical Manufacturing Systems

• Modern usage in representation reconstruction methods: an empirical way of GAN to provide solutions for multiple sectors  
  by Rohit Rastogi, Vineet Rawat, Sidhant Kaushal 
  Abstract: Image restoration poses a formidable challenge in the field of computer vision, endeavouring to restore high-quality images from degraded or corrupted versions. This research paper conducts a comprehensive comparison of three prominent image restoration methodologies: GFP GAN, DeOldify, and MIRNet. GFP GAN, featuring a specialised GAN architecture designed for image restoration tasks, introduces an AI-centric approach. DeOldify, a deep learning-based method, focuses on colourising and restoring old images using advanced AI techniques, while MIRNet offers a lightweight network specifically crafted for image restoration within an AI framework. The comparative analysis involves training and testing each method on a diverse dataset comprising both degraded and ground truth images. Employing a confusion matrix, precision, accuracy, recall, and other evaluation metrics are computed to comprehensively assess the performance of these AI-based methods. The matrix affords insights into the strengths and weaknesses of each AI-driven approach, providing a nuanced understanding of their respective performances.
  Keywords: GFP GAN; DeOldify; MIRNet; confusion matrix; restoration; image enhancement; computer vision; AI techniques; precision; accuracy; recall; noise; blur.