Floor cleaning plays a major role in all places, like homes, offices, etc., in the olden days by humans. Long-term cleaning makes the person get tired, and they cannot be involved in the deep and neat cleaning process. To make cleaning work easier and tidier, an automatic floor cleaning robot has been introduced through the AI technique. The digital camera is to make the autonomous system to navigate accordingly based on the environmental analysis. The main drawback of the system is that the robot automatically turns in another direction whenever it finds obstacles like different kinds of doors, poles of furniture, cables, small garments on the floor, etc. The purpose of the system is to move the autonomous robot move freely even it find the object to make the system to involve in deep cleaning. This document mainly proposes a vision based YOLOv5 framework to detect the object during navigation. The dataset is annotated from the scratch using different labels for 300 images. A novel approach proposed in this model is multi class object detection using the YOLOv5 and 6-DOF with the help of manual dataset. This work proposes a system that aims to develop a highly accurate automated robot system using the Computer Vision Annotation Tool (CVAT) and deep learning algorithm. The high-quality camera is fixed in front of the robot. Video is captured, and it is automatically converted into an image using the python script. The image is annotated manually using CVAT Tool which is processed using the deep learning technique of the YOLOv5 algorithm. The incorporation of obstacle avoidance capabilities prevents collisions with furniture and walls, contributing to a hassle-free cleaning experience with the concept of 6-DOF.This system makes the cleaning process efficient even when it finds obstacles, best model attained a mean average precision (mAP) of 93%.

In this paper, we compare various lidar point cloud merge algorithms for 3D map generation and use of high definition maps for autonomous vehicles (AV) localization. An autonomous vehicle may have a variety of sensors, including cameras, lidars and GPS sensors. Each sensor technology has its own pros and cons, for example GPS may not be very effective in a city environment with high-rise building; Cameras may not be very effective in poorly illuminated environments; and Lidars simply generate a relatively dense local point cloud. In a typical autonomous vehicle system, all of these sensors are present and sensor fusion algorithms are used to extract the most accurate information. By using our AV research vehicle, we drove on our university campus and recorded RTK-GPS (ZED-F9P) and Velodyne Lidar (VLP-16) data in a time synchronized fashion. In this paper, we focus on two different but related problems. The first one is, comparison of different point cloud merge algorithms for building the 3D map of the campus (a.k.a. high-definition map). The second one is a localization problem, given a high-definition map of the environment and a local point cloud data generated by a single lidar scan, determine the AV research vehicle's location. We will present a detailed analysis by using experimental data, and compare various merge and localization algorithms.

In the context of developing countries, medical instruments are imported from foreign countries. To overcome this challenge, the design of an oxygen analyzer using ultrasonic flow sensor technology and a microcontroller while promoting local innovation and reducing dependency on imported equipment is presented here. Moreover, it aims to enhance patient care by ensuring accurate oxygen concentration and flow rate measurements on ventilators and oxygen concentrators. The proposed system works well compared with the existing standard system. The measure data using the proposed system has been validated by comparing it with the data obtained using the standard oxygen analyzer equipment like VT-900 Gas Analyzer Fluke Biomedical and Ultra Max oxygen analyzer. Measurements were conducted on hospital ventilators with oxygen concentration which is also called fraction of inspired oxygen(FiO2) settings from 21% to 100% in 5% increments and flow rate settings from 1L/m to 10L/m. Results show an error value of 2.1% for oxygen concentration and 0.6 L/m for flow rate measurements. Based on the analysis, it can be concluded that the proposed system works well. Additionally, it offers portability, affordability, and user-friendliness, overcoming the limitations of existing options. This project seeks to contribute to the healthcare infrastructure in developing countries like Nepal, India, Bangladesh etc by providing a domestically produced solution for oxygen analysis.

Guided ultrasonic waves (GUW) analysis is a well-investigated method for structural health monitoring (SHM) applications. For plate-like structures, the pitch-catch technique is a popular choice since it offers the possibility to investigate a large area with a small number of sensors. This method requires a large amount of data to be analyzed to detect and localize damage. That, with the consequence that besides the presence of damage, also environmental influences like temperature and load will change the GUW signals. In addition, the location, size, and type of the damage will result in different changes of the GUW signals. Data-driven methods require sufficient data, requiring data augmentation. In order to get closer to this goal, this study aims to demonstrate the conversion of GUW signals measured with an air-coupled measurement system (ACMS) into signals measured with Piezoelectric Wafer Active Sensors (PWAS). This would allow the fast measurement of GUW data with ACMS at different positions of a plate-like specimen and translate it to a surface-bonded PWAS signal without the time-consuming process of transducer mounting. In this study, it is assumed that the measurement methods are not independent from each other when they are measured at the same position. To obtain the transfer function from ACMS to PWAS, GUW signals were measured both with ACMS and PWAS for different positions of artificial damage. Since both signal classes are physically dependent, it should be possible to determine the transfer function with machine learning (ML) methods. As input, the ACMS time-dependent signal or signal features are used, while the PWAS signals serve as labels for the training process. We are evaluating different ML-based transfer model architectures with respect to their suitability for signal or signal feature transformation, e.g., ANN, CNN, and LSTM-based networks, with a particular focus on Autoencoders.

This paper presents a novel approach to enhancing the safety of Advanced Driver Assistance Systems (ADAS) by integrating teleoperation for the remote control of ADAS features in a vehicle. The primary contribution of this research is the development and implementation of a teleoperation system that allows human operators to take control of the vehicle’s ADAS features, enabling timely intervention in critical situations where autonomous functions may be insufficient. While the concept of teleoperation has been explored in the literature, with several implementations focused on direct control of vehicles, there are relatively few examples of teleoperation systems designed specifically to utilize ADAS features. This research addresses this gap by exploring teleoperation as a supplementary mechanism that allows human intervention in critical driving situations, particularly where autonomous systems may encounter limitations. The teleoperation system was tested under two critical ADAS scenarios: cruise control and lane change assist, chosen for their importance in real-world driving conditions. These scenarios demonstrate how teleoperation can complement and enhance the performance of ADAS features. The experiments reveal the effectiveness of remote control in providing precise control, allowing for swift and accurate responses in scenarios where the autonomous system might face challenges. The novelty of this work lies in its application of teleoperation to ADAS features, offering a new perspective on how human intervention can enhance vehicle safety. The findings provide valuable insights into optimizing teleoperation for real-world driving scenarios. As a result of the experiments, it was demonstrated that integrating teleoperation with ADAS features offers a more reliable solution compared to standalone ADAS driving.

Wireless and battery-less sensor nodes are increasingly significant in advancing technology, particularly for continuous monitoring and reducing maintenance costs. These nodes are crucial in large-scale agricultural systems, where they facilitate essential tasks such as detection, identification, and fertilization by employing self-powered wireless sensors to ensure reliable and efficient performance. This paper investigate to enhance the power density and efficiency of electromagnetic energy harvesters by optimizing the performance within the system. According to Faraday’s law, the induced voltage is influenced by several key factors, including magnetic flux density, the number of coil turns, and the relative motion between magnets and coils. To thoroughly investigate these factors, three design configurations were modelled and analyzed under specific geometric constraints. The works involved using Finite Element Method Magnetics (FEMM) simulations to accurately measure magnetic flux density, followed by implementing MATLAB code to calculate the resulting voltage and power output for each design. The results demonstrated that the optimized magnetic arrangement led to a significant increase in both voltage and power output across all tested designs. Specifically, the best-performing configuration achieved an 20 % improvement in power output compared to the initial design. Furthermore, advanced mathematical techniques, including single-objective optimization was employed to further refine the power output which leading to enhance overall efficiency and performance of the energy harvester. This work will provide good insights for the high power energy harvester which use for the wireless sensor nodes.

Citrus farming in South Africa, has become extremely loop-sided in terms of economic
opportunities. The statistics shows that the wealthy, and large-scale farmers simultaneously
control 100% of the international export market and 77.1% of the local market hence,
endangering the prospect of the low and medium-scale farmers.

This research has developed a novel holistic conceptual design of a low-cost autonomous
mobile robot (AMR), capable of aiding small and medium-scale citrus farmers sustain
competitiveness. The AMR conceptual design was facilitated with the use of GENESYS
software for systems integration and architecting. The conceptual design of the AMR has
onboard system capabilities such as real-time monitoring of citrus plants on the field hence,
aiding small scale farmers comply with the stringent regulatory requirements of crop
monitoring and management.

Furthermore, the AMR conceptual design has provided for special features covering the
exhibition of autonomous navigation, detection of objects during navigation, alerting users
using a siren when issues are encountered along its path or when under a threat,
communicating errors when any is encountered and facilitation of auto-homing navigation
when the onboard battery reaches a 30% level of the total capacity. It is also GPS enabled and
possesses a robust onboard database system.

In addition, the AMR design is capable of periodic snapshotting of the citrus crops from the
sides, base and upper views. It utilises this capability to visually inspect the citrus plants for
infections hence, increasing their credibility for export and local markets. The system is also
capable of showcasing the protocols of crop nurturing especially as it affects the treatment
and recovery of infected crops. The novel AMR conceptual design, clustered with diverse
sensory systems, is a major shift in the deployment of low-cost smart technology in the citrus
farming industry for the small-medium farmers.

Stress analysis is crucial for understanding mental health and improving well-being. Electroencephalography (EEG) has recognised as a prominent tool for non-invasive stress detection due to its ability to capture brain signal. It records electrical activity of brain essential in early detection of stress. The research has suggested a number of Deep Learning (DL) techniques such as CNN, RNN, DNN, LSTM etc. for evaluating mental stress though there are various neuroimaging methods have been used to evaluate stress of human being. This review paper examines the recent advancements in stress analysis using EEG, focusing particularly on studies from the last five years that employ deep learning techniques. The review draws attention to the significant discrepancies among the research findings and makes the case that different data processing techniques lead to a number of contradicting conclusions. A number of factors, such as absence of a consistent protocol, the type of stressor, the brain area of interest, the duration of the experiment, appropriate EEG data analysis and the feature identification, extraction mechanism, and the type of classifier, could be responsible for the variances in the results. The incorporation of deep learning with EEG has shown significant potential in enhancing the precision accuracy and efficiency of stress identification systems.

Agriculture is the art of producing different crop types from the soil and plays an important role in our lives, sustaining and improving the economic sector. This study is mainly focused on spectral discrimination of crop types based on space-borne hyperspectral (PRISMA) sensor over Khanna, Amloh, Bassi Pathanan blocks lies in Punjab state, India. Hyperspectral sensor consists of narrow bands and provide precise, continuous spectral signature which can significantly help to obtain an unambiguous distinction among the crop types. Along with the reflectance and reflectance ratios, these attributes are useful for crop type discrimination using different combinations of metrics and classifiers. PRISMA hyperspectral sensor is used for spectral development and the collected end-member spectra of same crop types at different sites over study area were averaged to produce reference spectra for various specimens. This study evaluates the spectral discrimination between maize, sunflower, moong, sugarcane and chilli. A total of 135’ individual points are surveyed, and each collected field data was accompanied with photo record. Field data collection sites are selected by visual inspection of crop types present in the imagery data covering the study area. The highest spectral reflectance was shown in infrared spectral zone (940 to 1300 nm), relatively low reflectance in the spectral zone (1361 to 1449 nm and 1822 to 1932 nm) while the lowest reflectance was found in the spectral zone (2350 to 2495 nm). The visible region of the crop reflectance spectrum is characterized by low reflectance due to strong absorption by pigments like chlorophyll The implementation of new remote sensing technology in sustainable agriculture can be used more effectively for effective mapping, monitoring, post-harvest productions, minimizing the wastage and simplifying the transportation of output products etc.

The state of art Hybrid Brain Computer Interface (BCI) have shown improved classification of mental states either by combining different modalities or by choosing a combination of BCI activation tasks. Among these, the classification of motor imagery/executions of contralateral and ipsilateral data of upper arm is found challenging due to its spatial adjacency and retention of these spatial features. The proposed work uses a Hybrid BCI dataset acquired using EEG and fNIRS for upper limb movement (Right hand/ Left Hand, Right Arm/Left Arm). The electrode positioning is along the motor cortex and previous deep learning studies have shown that a good accuracy can be obtained without any channel selection. Hence the current study is to apply a combination of deep learning methods to the data which was halved into two without using channel selection algorithms. The model was evaluated for both set of channels using F1-score, Precision and Recall with an accuracy of 90%. This investigation shows that all the channels of the studied dataset contained inter-related spatial information. Also, the problem of long term EEG/fNIRS recording can be addressed using this study, if the total number of channels can be used in two halves by switching the channels after the minimum efficient time of recording.