2022· International Journal of Intelligent Automation & Robotics Engineering· Vol 5, pp. 01-08· 0 citations
TL;DR
The results show that the proposed framework is comparably better than classical embedded vision architectures in terms of object recognition, inference time, navigation accuracy and energy consumption.
Abstract
Intelligent embedded vision systems have emerged as a fundamental enabling technology for autonomous machines operating in industrial automation, autonomous vehicles, service robotics, agricultural robotics, healthcare systems, and smart surveillance. Unlike conventional machine vision systems that rely on centralized cloud processing, embedded vision integrates cameras, processors, artificial intelligence (AI), and edge computing into compact hardware platforms capable of performing real-time perception and decision-making. Now, the use of new advancements, such as deep learning, computer vision, edge AI accelerators and low-power embedded processors have vastly improved object detection accuracy, scene understanding and autonomous navigation at lower latencies and energy consumption. However, the embedded vision systems still have problems with locked in computational limitation, environmental variations, reliability of other sensors as well as cybersecurity and explainability of AI decisions. Abstract: This study provides a comprehensive overview and a conceptual framework for intelligent embedded vision systems of autonomous machines. As such, our architecture combines several components that include embedded cameras, AI on edge, sensor fusion, deep neural networks and adaptive control to optimize perception and operational intelligence. The results show that the proposed framework is comparably better than classical embedded vision architectures in terms of object recognition, inference time, navigation accuracy and energy consumption. This paper also pinpoints gaps and future opportunities with respect to current research including explainable AI, neuromorphic computer, federated learning and digital twin technologies. The results provide researchers and practitioners with insights into designing the next-generation autonomous robotic platforms to be ready for Industry 5.0 environment.
The rapid advancement of intelligent embedded robotic systems has significantly transformed industrial automation, healthcare, logistics, agriculture, surveillance, and autonomous transportation. The current vision for robotic platforms is one of fast perception, reasoning and control with strict computational and energy limits of operation. Communication latency, bandwidth bottleneck (the whole data is not always collected in the cloud), privacy and operational reliability loss are well-known issues of traditional cloud-centric architectures deployed for environments where network connectivity may be erratic. Recently, Edge computing have shown a potential to shift computational intelligence closer to robotic devices allowing for low latency processing[10], decentralized intelligence as well as superior autonomy. This paper explores edge computing architectures tailored for intelligent embedded robotic systems through an analysis of the architectural components, computational workflow, communication mechanisms, and performance characteristics. This article presents an overview of the state-of-the-art on edge-oriented robotics, with a focus on how artificial intelligence, Internet of Things (IoT), embedded systems and machine learning models are integrated in current edge environments, while providing context for some existing research challenges. Moreover, the intelligent perception for robot at edge (edge computing), local decision making faculty, collaborative processing on edge and cloud-assisted learning framework will be discussed based on layered architecture. A comparative analysis shows that edge-based robotic systems enhance responsiveness, network efficiency, scalability, and operational reliability as compared to traditional cloud-dependent systems. Emerging trends such as federated learning, digital twins, adaptive edge orchestration and 6G-enabled robotic ecosystems are also discussed. The results demonstrate edge computing as a true enabled technology for the future of autonomous embedded robotic systems to operate smartly, securely and energy-efficiently in dynamic real-world environments.
Alan Bundy, Karen Spärck Jones· International Journal of Int...· 0 citations
This study presents a comprehensive research framework for real-time embedded AI-based industrial robot monitoring that combines edge computing, embedded deep learning, multi-sensor fusion, anomaly detection, predictive maintenance, and intelligent decision-making and provides a scalable, energy-efficient, and intelligent monitoring solution suitable for next-generation smart factories and Industry 5.0 environments.
Corrado Böhm, Corrado Gini· International Journal of Int...· 0 citations
Autonomous robotic platforms have become an integral part of industrial automation, intelligent manufacturing, autonomous transportation, precision agriculture, healthcare robotics, and hazardous environment exploration. The operational reliability of these robotic systems strongly depends on the accuracy and health of their sensing infrastructure. Sensors including inertial measurement units (IMUs), LiDAR, cameras, ultrasonic sensors, GPS modules, encoders, force-torque sensors, and proximity sensors continuously provide environmental and operational information for autonomous decision-making. However, sensor degradation, calibration drift, communication failures, environmental interference, and hardware aging can significantly deteriorate robotic performance and may even lead to catastrophic failures. Consequently, intelligent sensor fault diagnosis has emerged as a critical research area that combines artificial intelligence, machine learning, data analytics, and model-based reasoning to identify, classify, and predict sensor faults in real time. In this paper, a complete intelligent sensor fault diagnosis framework for autonomous robotic platforms is proposed. The proposed framework utilizes the integration of multi-sensor data fusion, feature extraction, deep learning-based fault classification, anomaly detection, and predictive maintenance components into a unified architecture. Experimental performance evaluation shows significant enhancements versus traditional threshold-based diagnostic approaches in terms of high fault detection accuracy, lower false alarms and reduction in the latency for diagnosis among different faults with an increased reliability level on system diagnosis. These results demonstrate that AI-assisted diagnostic models significantly increased robotic autonomy through the ability to proactively manage faults, reduce downtime and enhance operational safety. The proposed framework is suitable for deployment in Industry 5.0 manufacturing systems, autonomous vehicles, collaborative robots, and intelligent service robotics.
James Carter, Patricia Hall· International Journal of Int...· 0 citations
Experimental evaluation demonstrates significant improvements in inspection accuracy, fault classification, decision-making, maintenance prediction, reduced downtime, and lower human intervention, providing a foundation for next-generation intelligent industrial automation.
O. Dahl, K. Nygaard· International Journal of Mod...· 0 citations
Robotic object recognition is a fundamental capability that enables autonomous robots to interact intelligently with dynamic environments. Traditional vision-based methods, such as SIFT, SURF, HOG, and template matching, perform well under controlled conditions but struggle with variations in lighting, viewpoint, occlusion, and complex backgrounds. Recent advances in deep learning have significantly improved recognition accuracy by automatically learning features from raw image data. However, individual deep learning models often face challenges related to computational cost, inference speed, and limited generalization. This paper proposes a Hybrid Deep Learning Framework that integrates Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), attention mechanisms, and multimodal sensor fusion (RGB, depth, and LiDAR) to enhance recognition accuracy and efficiency. The framework combines local and global feature extraction, adaptive feature fusion, intelligent object recognition, and robotic decision-making for real-time perception and task execution. It supports applications in industrial automation, warehouse logistics, autonomous mobile robots, healthcare, agriculture, and service robotics while improving robustness, scalability, and computational efficiency.
Seshagiri N, Mahabala H. N.· International Journal of Int...· 0 citations
The rapid evolution of intelligent robotic systems has increased the demand for embedded computing architectures capable of performing complex perception, decision-making, and control operations under strict energy constraints. Traditional robotic platforms often rely on centralized cloud computing or high-performance processors, resulting in increased communication latency, energy consumption, and reduced autonomy. Energy-aware embedded intelligence overcomes these drawbacks by directly embedding lightweight artificial intelligence (AI) algorithms, hardware accelerators, adaptive power management and edge-based decision-making capabilities into robotic platforms. We propose an energy-smart embedded intelligence framework for a smart robotic application, engaging the ensemble of (i) embedded AI (ii) sensor fusion mechanism (iii), Nature-mimicry based energy optimization strategy and (iv) real-time autonomous control. We have designed robotics based on dynamic workload scheduling and intelligent resource allocation for dynamically balancing the computational performance with power consumption. The paper introduces the results of an analysis on state-of-the-art in embedded intelligence methods, short description of research gaps and also insight regarding energy-efficient architectures effectiveness that addresses mobile robots, autonomous robotic systems, and more broadly industrial robots. Comparative evaluation shows that energy-aware embedded intelligence can drastically reduce energy consumption and maintain high accuracy and responsiveness. It suggests that enhanced edge-intelligence architectures will drive sustainable, autonomous, and scalable operation of next-generation robotic systems in Industry 5.0 settings.
Narendra Karmarkar· International Journal of Int...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
