Journal of Artificial Intelligence, Machine Learning and Soft Computing

Abstract

One of the most important tasks of an autonomous system of any kind is to acquire knowledge about its environment. This is done by taking measurements using various sensors and then extracting meaningful information from those measurements. In this paper, we present the most common sensors used in mobile robots, classify them, and then discuss strategies for extracting information from the sensors. Sensors can be classified on the basis of two important functional categories: proprioceptive/exteroceptive and passive/active. Proprioceptive sensors are involved in measurement of values that are internal to the system (robot); such as motor speed, wheel load, robot arm joint angles, battery voltage. Exteroceptive sensors get information from a robot’s environment; including, sound amplitude, intensity of light, distance measurements, etc

Keywords: Sensors in Autonomous Systems, Multi-sensor Systems, Exteroceptive Sensors

Authors: Vijay Khare, Deepak Singhal

Abstract: This Paper demonstrate the performance comparison of two artificial neural network (ANN) base techniques Levenberg-Marquardt and Radial basis function network for classification of planning of right hand movement with respect to an awake relax state. In this study data was collected from 10 healthy subjects. Wavelet packet transform (WPT) was used for Feature extraction of the relevant electroencephalogram (EEG) signals.

Abstract

The Visual Awareness Kit (VAK) is an artificial intelligent system that helps the visually impaired people in the society. The proposed system is made with an innovative idea to help the blind to navigate easily through obstacles. A blind man has so many limitations while walking through the street, to identify an object, to avoid obstacles in front of him/her etc.  As mentioned, VAK helps the blind, by identifying the scenes in front of him/her and the distance to the nearest obstacle. The information includes what exactly the obstacle is and the distance between him/her with the obstacle. And this information is communicated to the person through a headset. All this is done using artificial neural networks. The VAK consists of an autofocus camera for live image capturing, A high performance Raspberry pi module trained with convolution neural network for processing and captioning the image input from the camera..

Keywords: Visual Awareness Kit (VAK), Raspberry pi, artificial neural networks.

Abstract

The integration of Artificial Intelligence (AI) into robotics has revolutionized the capabilities of machines, enabling them to perform complex tasks autonomously in dynamic and uncertain environments. This paper explores the multifaceted domain of robotics and autonomous systems with an emphasis on AI-driven approaches to motion planning, sensor fusion, and real-time decision-making. It delves into how intelligent algorithms empower robots to navigate, adapt, and execute tasks without human intervention. The discussion highlights core advancements in perception systems, control algorithms, deep learning, and reinforcement learning, which together enable autonomous behaviors. This paper also reviews challenges such as environmental uncertainty, safety, and explainability in autonomous actions, offering insights into future trends and applications in sectors like transportation, manufacturing, healthcare, and defense.

Keywords: Robotics, Autonomous Systems, Artificial Intelligence, Motion Planning, Sensor Integration, Real-Time Decision-Making, Perception, Machine Learning, Autonomous Navigation

Abstract

This paper presents a comprehensive study on intelligent agents and multi-agent systems (MAS), emphasizing how multiple autonomous entities can interact, coordinate, negotiate, and solve complex problems collaboratively. As technology moves towards decentralization and distributed intelligence, MAS offers robust frameworks for achieving scalable, adaptive, and intelligent behavior in fields such as robotics, smart grid management, healthcare, and simulation. This paper explores the fundamental architecture of intelligent agents, their decision-making processes, learning abilities, and the underlying mechanisms that allow effective multi-agent interaction. The discussion includes negotiation models, coordination strategies, communication protocols, and conflict resolution methods. Furthermore, this work highlights real-world applications and future directions for MAS in evolving technology landscapes.

 Keywords: Intelligent Agents, Multi-Agent Systems, Autonomous Agents, Coordination, Negotiation, Distributed AI, Swarm Robotics, Agent-Based Simulation, MAS Applications

Abstract

Computer Vision is a transformative field of artificial intelligence that enables machines to understand and interpret visual information from the real world. This paper explores the fundamental techniques used in image and video recognition, object detection, and scene understanding. It delves into the algorithms and deep learning architectures that power these systems and highlights their practical applications in sectors such as autonomous vehicles and medical imaging. Through a comprehensive analysis of current methodologies, challenges, and emerging trends, the paper aims to present a clear understanding of the trajectory of computer vision and its impact on technological innovation.

Keywords: Computer Vision, Image Recognition, Object Detection, Scene Understanding, Deep Learning, Convolutional Neural Networks, Autonomous Vehicles, Medical Imaging, AI in Healthcare, Machine Perception

Abstract

Natural Language Processing (NLP) represents a dynamic field of Artificial Intelligence (AI) that bridges the gap between human communication and machine interpretation. This paper provides a comprehensive exploration of the evolution, methodologies, and applications of NLP, with an emphasis on sentiment analysis, machine translation, and question-answering systems. It discusses the transition from traditional rule-based and statistical models to neural network-based architectures and the recent emergence of large language models (LLMs). The research further examines the social and ethical implications of LLMs in real-world applications and their transformative impact across industries. With practical illustrations, original figures, and comparative tables, the paper offers a deep dive into the capabilities, limitations, and future trajectory of NLP technologies.

Keywords: Natural Language Processing, Sentiment Analysis, Machine Translation, Question Answering, Large Language Models, Transformer Models, BERT, GPT, Neural Networks, NLP Applications

Abstract

The paper presents a detailed exploration of machine learning (ML) and deep learning (DL), focusing on the theoretical foundations, core algorithms, and practical applications across diverse industries. It delves into the fundamental categories of machine learning, namely supervised, unsupervised, and reinforcement learning, highlighting key algorithms, working principles, and use cases. In parallel, the paper investigates the architecture and application of deep learning models such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), showcasing their transformative role in image recognition, natural language processing, time series analysis, and autonomous systems. Through comparative analysis and visualization, the paper bridges theoretical constructs with domain-specific implementations, offering valuable insights into how intelligent systems can be developed using modern data-driven approaches.

Keywords: Machine Learning, Deep Learning, Supervised Learning, Unsupervised Learning, Reinforcement Learning, CNN, RNN, Neural Networks, Data-Driven Models, Artificial Intelligence

Abstract

Speech and audio processing has become a cornerstone in modern human-computer interaction, enabling systems to understand, analyze, and generate audio signals. Applications span voice assistants, automatic speech recognition (ASR), speaker identification, audio event detection, and music analysis. This paper provides a comprehensive review of contemporary speech and audio processing algorithms, examining their foundations, advancements, and practical applications. Traditional signal processing methods, such as Fourier transforms and filter banks, are discussed alongside modern machine learning and deep learning approaches including convolutional neural networks (CNNs), recurrent neural networks (RNNs), and transformers. Additionally, challenges such as noise robustness, real-time processing, and resource-efficient deployment are highlighted. The review also presents comparative studies, illustrative figures, and performance metrics of different algorithms.

Keywords: Speech Processing, Audio Signal Processing, Machine Learning, Deep Learning, Fourier Transform, Spectrogram, Automatic Speech Recognition, Speaker Identification

Abstract

Representation learning has become one of the most important research directions in modern Artificial Intelligence. Instead of relying heavily on large amounts of labeled data, self-supervised and unsupervised learning techniques aim to learn meaningful data representations from unlabeled data. These approaches reduce the dependency on human annotations and allow models to learn hidden structures and patterns present in raw data. Self-supervised learning creates surrogate tasks from the data itself, while unsupervised learning extracts intrinsic patterns without labels. This paper reviews major techniques, architectures, and recent developments in self-supervised and unsupervised representation learning across vision, natural language processing, and speech domains. Methods such as autoencoders, contrastive learning, generative models, masked modeling, and clustering-based learning are discussed in detail. Applications in healthcare, robotics, recommender systems, and autonomous systems are highlighted. Challenges and future directions are also examined.

Keywords: Self-supervised learning, Unsupervised learning, Representation learning, Contrastive learning, Autoencoders, Masked modeling, Generative models, Deep learning

Abstract

Machine learning (ML) has revolutionized various sectors including healthcare, finance, autonomous systems, and cybersecurity. However, real-world deployment exposes models to uncertainties, noisy inputs, and malicious adversarial attacks. Robust machine learning focuses on enhancing model stability against noise and perturbations, while adversarial machine learning studies deliberate attacks that exploit model vulnerabilities. This paper provides a comprehensive review of robust and adversarial ML, covering recent advancements, threat models, attack and defense mechanisms, evaluation metrics, and practical applications. Additionally, we explore the challenges of achieving both robustness and accuracy and highlight future research directions. A detailed discussion of defense strategies, including adversarial training, certified robustness, and robust optimization, is presented, along with comparisons of current methodologies.

Keywords: Robust Machine Learning, Adversarial Attacks, Adversarial Training, Certified Robustness, Deep Learning Security, Threat Models, Perturbation Analysis

Abstract

In this paper, prototype of an energy harvesting mobile robot is presented which harvests energy as soon as its power reserves diminish below a predecided level. The need for restoring power in the prototype mobile robot drives it to auto plug-in with a microcontroller interface of a motor control unit. Development of the system involves making a designing of various parts of a prototype mobile robot. The mobile robot at first follows a line which is nothing but a line following the robot. With the movement of the robot, the battery charge will be decreased. When the charge decreases to less than 50%, the mobile robot then follow the second line to recharge itself by auto plugging into the charging station. After getting fully recharged, the mobile robot again switched to follow the first line. A control circuit is designed to control the line following movement of the mobile robot. An algorithm is also developed that command the control circuit. The control circuit consists of infrared sensor, motor driver circuit with power supply, transmitter circuit, receiver circuit and infrared LED (IR-LED). DC motor is used as an actuator to control the wheel of the mobile robot. An infrared sensor is used to generate high and low frequency in the transmission circuit. High frequency is generated when capacitor’s capacity is low and low frequency is generated when capacitor’s capacity is high. The receiver circuit receives the high and low frequency and sends signal to the program that controls the DC motor according to the analysis result. Thus, the DC motor drives the wheel to control the movement of the mobile robot. The overall success rate of the prototype mobile robot is 83.33% including line following and recharging.

Keywords: Intelligent energy harvesting, Auto-recharging, Emergent situation, Recharging, Capacito

Abstract

Many countries of the world are facing a big problem of the terrorism nowadays. Terrorism is of many kinds but the main reason for the resulting violence are not the same. Soldiers of the countries affected by terrorism have increased responsibilities on them to protect citizens. The Indian government faces the same problems as any other government fighting cross border terrorism. There are many uneven grounds on which terrorists operate in India, where it is difficult for the soldiers to reach.  One of the solutions to this problem is to develop a robotic system which will help the soldiers to reach those difficult to reach places.

Many techniques have been developed in the robotics, but still they are not very effective to detect terrorist activities. This paper gives an overview of the past few techniques in this field and presents a real–time system by which a robot will follow objects to capture videos and transmit them to a server. This technique is based on the Papillion Spartan3 FPGA in which Android is used by the soldiers for viewing live videos.

Keywords: Papilio, Spartan3, FPGA, Android

Abstract

Companies have been collecting data for decades, and building massive warehouses for data storage. Even though this data is available, very few companies have been able to realize its actual value. The question these companies are asking is how to extract this value. For this, there are many technologies available to Soft Computing practitioners, including Artificial Neural Networks, Regression, and Decision Trees. Many practitioners are wary of Neural Networks due to their black box nature, even though they have proven themselves in many situations. This paper is an overview of artificial neural networks in computer industry with respect to soft computing.

Keywords: Artificial Neural Network (ANN), Neural Network Topology, Advantages and Applications.

Abstract

Reinforcement Learning (RL) has emerged as a powerful paradigm in artificial intelligence, enabling agents to learn optimal behaviors through interaction with dynamic environments. Traditional RL focuses on single-agent scenarios where an agent maximizes cumulative rewards through trial-and-error learning. However, real-world applications often involve multiple interacting agents, leading to the development of Multi-Agent Reinforcement Learning (MARL). MARL introduces challenges such as non-stationarity, coordination, and scalability, demanding innovative solutions and learning frameworks. This paper provides a comprehensive review of RL and MARL, discussing fundamental principles, key algorithms, theoretical foundations, practical applications, and ongoing challenges. Furthermore, it examines recent advancements in MARL strategies, including decentralized learning, cooperative-competitive frameworks, and emergent behaviors. The paper also highlights promising applications across autonomous systems, robotics, smart grids, and traffic management, emphasizing the transformative potential of RL and MARL in complex, dynamic environments.

Keywords: Reinforcement Learning, Multi-Agent Reinforcement Learning, Q-Learning, Policy Gradient, Deep RL, Cooperative Agents, Competitive Agents, Game Theory, Autonomous Systems

Abstract

Quantum Machine Learning (QML) is an emerging interdisciplinary field that combines principles of quantum computing with classical machine learning techniques to solve complex computational problems more efficiently. With quantum algorithms promising exponential speed-ups for certain tasks, QML has the potential to revolutionize domains ranging from optimization and cryptography to drug discovery and artificial intelligence. This paper provides a comprehensive review of QML, exploring its foundational concepts, algorithmic advancements, hybrid quantum-classical architectures, and potential applications. Furthermore, the paper discusses current challenges, such as noise in quantum systems, scalability, and resource limitations, along with future research directions. Illustrative tables and figures are included to clarify the workflow and advantages of QML over classical methods.

Keywords: Quantum Machine Learning, Quantum Computing, Hybrid Algorithms, Variational Quantum Circuits, Quantum Neural Networks, Quantum Optimization

Abstract

Neurosymbolic Artificial Intelligence (AI) represents a promising paradigm that bridges the gap between neural networks and symbolic reasoning. Traditional neural methods excel at pattern recognition from large datasets but often lack explainability and reasoning capabilities. Conversely, symbolic AI offers robust logical reasoning but struggles with unstructured data and learning from examples. Neurosymbolic AI integrates these paradigms, combining neural models' adaptability with symbolic systems' interpretability and knowledge representation. This paper reviews the state-of-the-art in neurosymbolic AI, covering its conceptual framework, modeling techniques, applications across domains, challenges, and future directions. The integration of neural and symbolic methods offers a path toward more robust, interpretable, and generalizable AI systems, with potential implications for healthcare, robotics, natural language understanding, and autonomous systems.

Keywords: Neurosymbolic AI, Neural Networks, Symbolic Reasoning, Knowledge Representation, Explainable AI, Hybrid Intelligence

Abstract

Robotics is the future. And the future is now. Robotics was first used in automation of large scale industries and then small scale industries. Now, robotics is being introduced into fields where there is a scope for reduction of human effort. Households of today are becoming smarter and more automated. Home automation delivers convenience and frees up more time for people. Domestic robots are entering the homes and people’s daily lives, but it is still a relatively new and immature market. However, growth is predicted and adoption of domestic robots is evolving. Several robotic vacuum cleaners are available on the market but only a few specialize in wiping of floors with water. This paper presents a project that follows a product development process, includes the stages of project planning, market research, setting a requirement specification, concept building, concept evaluation and selection and embodiment design. It entails our journey to creating a compact yet powerful waste management robot that is capable of performing both wet and dry cleaning operations. The idea is to build a robotic device that sweeps while traversing the whole enclosed area in one phase and then wipes the area in the other. The constructed prototype serves as a learning tool that facilitates in revealing new aspects and recognizing features that should be developed in the robot. A technology-intensive product of this kind requires the end users to adapt to a new technology. Therefore, a user-centred approach is strongly recommended to ensure better marketability of the product.

Keywords: Robotics, Automation, Arduino