Scholarlink

<p>Industry 4.0, also known as the Fourth Industrial Revolution, has transformed the manufacturing sector by integrating intelligent technologies such as artificial intelligence (AI), cloud connectivity, and real-time data analysis. The objective is to enhance efficiency and productivity, and to enable the development of cyber-physical systems. Building on this momentum, Industry 5.0 further reinforces the digital transformation by encouraging closer collaboration between humans and intelligent machines. Digital twins have recently emerged and revolutionized industrial production environments by creating highly accurate virtual replicas of their physical counterparts. A digital twin is a continuously updated model that closely mirrors its physical counterpart with real-time data, and can accurately predict its future states. For production equipment, the adjustment of the digital twin to its physical counterpart poses a considerable challenge during commissioning, and remains a significant obstacle during the operating phase. The current approach relies on the use of the Industrial Internet of Things (IoT) to enhance monitoring. This approach requires excessive instrumentation of the physical system, leading to complications in terms of cost and reliability. Over time, machine vision has established itself as a mature and powerful technology, enabling the integration of 3D machine learning-based recognition solutions. However, its applications remain primarily focused on product monitoring. Sensor-based calibration methods including IMU, LiDAR and multi-sensor-fusion provides a fit solution for spatial alignment but stumbles with adaptability and real-time accuracy. On the other hand, Vision-Based calibration, mainly neuromorphic cameras and system vision techniques, is more dynamic, provides with self-adjusting solutions with high precision. This Literature Review classifies and critically evaluates the calibration technologies, highlighting their challenges, application and recent innovations. We analyzed 35 articles selected from a total of 447, and We present a machine vision architecture for calibrating and synchronising a digital twin of production equipment. This architecture is based on the automatic generation of behavioural models using discrete event systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Towards accurate digital twins: A review of calibration methods using event-based and neuromorphic vision

  • 期刊论文

  • 发表日期：

    2026年5月1日

查看全文
  • Jesus Vital Kombaya Touckia,
  • Sébastien Henry

摘要

Industry 4.0, also known as the Fourth Industrial Revolution, has transformed the manufacturing sector by integrating intelligent technologies such as artificial intelligence (AI), cloud connectivity, and real-time data analysis. The objective is to enhance efficiency and productivity, and to enable the development of cyber-physical systems. Building on this momentum, Industry 5.0 further reinforces the digital transformation by encouraging closer collaboration between humans and intelligent machines. Digital twins have recently emerged and revolutionized industrial production environments by creating highly accurate virtual replicas of their physical counterparts. A digital twin is a continuously updated model that closely mirrors its physical counterpart with real-time data, and can accurately predict its future states. For production equipment, the adjustment of the digital twin to its physical counterpart poses a considerable challenge during commissioning, and remains a significant obstacle during the operating phase. The current approach relies on the use of the Industrial Internet of Things (IoT) to enhance monitoring. This approach requires excessive instrumentation of the physical system, leading to complications in terms of cost and reliability. Over time, machine vision has established itself as a mature and powerful technology, enabling the integration of 3D machine learning-based recognition solutions. However, its applications remain primarily focused on product monitoring. Sensor-based calibration methods including IMU, LiDAR and multi-sensor-fusion provides a fit solution for spatial alignment but stumbles with adaptability and real-time accuracy. On the other hand, Vision-Based calibration, mainly neuromorphic cameras and system vision techniques, is more dynamic, provides with self-adjusting solutions with high precision. This Literature Review classifies and critically evaluates the calibration technologies, highlighting their challenges, application and recent innovations. We analyzed 35 articles selected from a total of 447, and We present a machine vision architecture for calibrating and synchronising a digital twin of production equipment. This architecture is based on the automatic generation of behavioural models using discrete event systems.