machine Related Articles

Machine vision cameras in 2025 use advanced sensors and artificial intelligence to capture and process images with high speed and accuracy. These cameras support automation, robotics, and computer vision systems by enabling real-time inspection and precise measurement. Recent trends highlight the use of edge AI and synthetic data, which help companies reduce costs and speed up operations. Market reports show strong growth, with machine vision becoming essential for industries like automotive and electronics. The rapid adoption of these cameras demonstrates their importance for quality control and efficient production.Machine Vision Cameras OverviewWhat Are Machine Vision CamerasMachine vision cameras help machines see and understand their surroundings. These cameras use advanced sensors and computer vision to capture images and send them to a processing system. The system analyzes the images to find objects, measure parts, or check for defects. In 2025, machine vision cameras work in many industries, such as automotive, electronics, food, and healthcare. They support tasks like quality assurance, inspection, and position guidance.The global market for machine vision continues to grow. In 2024, the market size reached over $20 billion. Experts expect it to double by 2030. Asia-Pacific leads with more than 43% of the market share. Hardware, including cameras and sensors, makes up the largest part of this market. Companies like Cognex, Basler, and Teledyne focus on research and development to improve these products.Metric CategoryDetails / Quantitative DataMarket SegmentationArea scan, line scan, 3D cameras; CMOS vs CCD sensors; USB, Ethernet, Camera Link, CoaXPressRegional GrowthAsia Pacific CAGR: 10.2%; North America and Europe: steady, mature marketsSensor Technology ShiftCMOS preferred for lower power, faster frame rates, better integrationApplication SectorsAutomotive, electronics, food & packaging, pharmaceuticalsProduct Type UsageArea scan: versatile; Line scan: high-speed; 3D cameras: depth perceptionConnectivity TrendsUSB 3.0 for cost; Ethernet for long-distance, high-speed transferMarket DriversIndustrial automation, Industry 4.0, smart factories, quality control, sustainabilityCompetitive LandscapeMajor players: Cognex, Basler, Teledyne, FLIR; focus on R&D and product expansionImage Source: statics.mylandingpages.coKey Features and DifferencesModern machine vision cameras stand out from regular cameras because they offer high accuracy, speed, and reliability. These cameras use AI and machine learning to recognize objects and detect defects. They process images quickly, which helps factories speed up production and reduce errors. Many cameras now use edge computing, so they can analyze images right where they are captured.Note: Machine vision cameras can detect defects with over 99% accuracy and improve defect detection rates by up to 30% compared to older systems.The table below shows how machine vision cameras compare to traditional imaging systems:Feature / MetricDescription / ValueAccuracyOver 99%, industry benchmark at 99.8%Recall100%Precision99.6%SpeedFaster inspection and production cyclesEfficiencyStreamlined quality controlCost SavingsLess waste, rework, and downtimeDefect Detection RateUp to 30% improvementMaintenance Savings20–40% reductionProduction Efficiency20–30% increaseSafety Incidents40–60% reductionDowntime30–50% reductionMachine vision cameras also support different types of sensors, such as CMOS and CCD, and can work with various platforms, including smart cameras and PC-based systems. They adapt to many environments and connect easily to other machines using USB or Ethernet. These features make machine vision a key part of modern automation and computer vision solutions.Core Components of Machine Vision SystemsImage Source: pexelsImage Sensors and OpticsMachine vision systems rely on advanced image sensors and optics to capture high-resolution images. The sensors in industrial cameras measure light and convert it into digital signals. These signals allow vision systems to detect fine details, even in challenging environments. The choice of sensor and lens affects the accuracy and reliability of machine vision. Different applications require different sensor sizes, lens mounts, and focal lengths. For example, a larger sensor paired with the right lens mount prevents issues like stray light and shading. This careful pairing ensures that machine vision systems deliver consistent results.The table below highlights key performance metrics and optical factors that impact machine vision systems:Performance Metric / Optical FactorExplanation / Impact on Machine Vision SystemsFrame rate, dynamic range, resolution, pixel size, shutter type, ROI and binning output modesThese metrics determine the sensor's ability to capture accurate, high-quality images under dynamic and challenging lighting conditions, which is critical for machine vision accuracy and reliability.Sensor size and optical formatInfluence lens compatibility and image quality; larger sensors require appropriate mounts and optics to avoid performance degradation.Lens mount compatibility (e.g., C-Mount, TFL mount)Proper pairing of mount to sensor size avoids optical issues such as shading, stray light, and mechanical interference, enhancing system robustness.Focal length to sensor size ratioRecommended ratio of 2:1 to 4:1 ensures lenses can achieve required optical performance, especially for high-resolution sensors; ratios near or below 1:1 cause performance challenges.Lens distortion and optical design tradeoffsDifferent lenses perform variably at different working distances; no single lens fits all applications, so selection based on application needs is essential.Note: Matching optics to sensor size and application needs helps machine vision systems maintain durability and accuracy in industrial settings.Lighting and InterfacesLighting plays a vital role in machine vision systems. Proper lighting ensures that cameras capture clear and consistent images. Industrial cameras often use synchronized lighting, which works with hardware triggers and GPIO interfaces. This setup allows vision systems to respond within microseconds, making inspections fast and reliable. Bit depth, ranging from 8-bit to 16-bit, helps cameras handle different lighting conditions and maintain image quality.Modern machine vision systems use several interface standards. USB3 Vision, GigE Vision, CoaXPress, and Camera Link each offer unique benefits. For example, GigE Vision supports long cable runs and multi-camera setups, which helps factories scale their vision systems. CoaXPress provides high bandwidth for applications that need fast or high-resolution images. These interfaces make it easier to integrate machine vision into smart factories. New standards, such as OPC UA Companion Specifications, help different vision systems communicate and work together. This standardization supports efficiency and reliability in Industry 4.0 environments.How Machine Vision WorksImage Capture ProcessMachine vision systems begin their work with the image capture process. Cameras in these systems use advanced sensors to collect visual data from their environment. The sensors detect light and convert it into digital signals. These signals form the raw images that vision systems need for further analysis. Industrial image processing depends on the quality of this first step. High-resolution sensors and precise optics help capture clear and detailed images, even in fast-moving production lines.Lighting plays a key role in this process. Proper lighting ensures that cameras can see every detail, which is critical for industrial image processing. Many machine vision systems use synchronized lighting and hardware triggers. This setup allows cameras to capture images at the exact moment needed. The result is a sharp, consistent image that supports accurate inspection.A recent scientific study highlights the efficiency of modern machine vision systems. Researchers combined a calibrated camera with the YOLOv5 deep learning model. The system achieved a mean Average Precision (mAP) of 0.95 and an error rate of only 0.2%. These results show that machine vision cameras can detect features with high accuracy and minimal errors. This level of performance surpasses many industry standards and supports reliable automated image production.Tip: Consistent lighting and high-quality sensors help machine vision systems maintain accuracy and efficiency in demanding environments.Processing and AnalysisAfter capturing images, machine vision systems move to the processing and analysis stage. Here, image processing software takes the digital signals and extracts useful information. The software uses algorithms to find objects, measure parts, and check for defects. Industrial image processing relies on these steps to ensure products meet quality standards.Modern machine vision systems use advanced techniques to speed up analysis. Deep learning models, such as YOLOv5, help systems recognize complex patterns and features. Feature selection methods, like Feature variable Dimensional Coordination (FDC), reduce the amount of data the system needs to process. This approach cuts processing time by up to 61% while keeping accuracy above 90%. Other methods, such as Random Forest, can complete analysis in just 11 seconds and maintain accuracy above 95%.The table below shows how different techniques impact processing time and accuracy in machine vision systems:Technique / MethodProcessing Time / ReductionAccuracy / Performance MetricsFeature variable Dimensional Coordination (FDC)Reduced processing time by up to 61%Maintained accuracy above 90%Random Forest11 seconds (fast)Accuracy above 95%Hybrid V-WSP-PSO Feature SelectionReduced features from 27,620 to 114High predictive performance (RCV2 = 0.9908)Frequency-based Tokenization (1-mer)N/AAccuracy ~97%Machine vision systems often use embedded processors or edge devices for real-time analysis. These devices run image processing software directly on the camera or nearby hardware. This setup reduces delays and supports fast decision-making in industrial settings. Computer vision and industrial image processing work together to help factories spot defects, guide robots, and improve safety.Note: Advanced image processing software and hardware integration allow machine vision systems to deliver fast, accurate results in real-world applications.Machine vision continues to evolve with new algorithms and hardware. These advancements make vision systems more reliable and flexible. Industrial image processing now supports a wide range of tasks, from quality control to automated image production. As a result, machine vision systems have become essential tools in modern factories and smart manufacturing environments.3D Machine Vision Camera TechnologyImage Source: pexels2D vs 3D Machine VisionA 3d machine vision camera captures depth and shape, while a 2D system only records flat images. This difference allows 3d machine vision camera systems to measure objects in three dimensions, making them ideal for tasks that require precise geometry. Researchers compared 2D and 3d machine vision camera systems using both static and dynamic calibration. They found that static calibration, which uses a single image, produced a mean error of about 0.4 mm. Dynamic calibration, which uses many frames over time, had a slightly higher error of 1.3 mm. Professional-grade 3d machine vision camera models showed less than 0.4 degrees of error, while consumer-grade cameras performed nearly as well.The study also showed that 2D systems are more sensitive to lighting and reflections. Overexposure and blurred edges can reduce accuracy in 2D images. In contrast, a 3d machine vision camera excels at measuring depth and geometry, but curved surfaces can still present challenges. Combining both 2D and 3d machine vision camera data led to over 95% true positive accuracy and reduced inspection time by 20% compared to human inspection.Tip: Using both 2D and 3d machine vision camera data together can improve accuracy and speed in industrial inspections.Applications of 3D CamerasA 3d machine vision camera finds many uses in modern industry. Thousands of scans from different 3d machine vision camera devices help engineers model noise and improve data quality. This process trains neural networks to perform better on real-world tasks. High-resolution 3d machine vision camera images allow for precise measurements, such as checking surface texture or detecting small defects.Many factories use 3d machine vision camera systems for flow visualization, fault detection, and process monitoring. Event-based sensor technology in these cameras can reach frame rates up to 10,000 fps, much faster than traditional cameras. This speed helps detect small changes and high-frequency movements, such as vibrations or acoustic signals. Neuromorphic sensors in some 3d machine vision camera models work well even in poor lighting, handling reflections and shadows better than standard cameras.A 3d machine vision camera also reduces costs by minimizing image retakes and centralizing data collection. Pre-calibrated settings ensure consistent image quality, which is important for quality assurance. These features make the 3d machine vision camera a key tool for modern machine vision applications.Applications in 2025Industrial AutomationFactories in 2025 rely on machine vision cameras to boost productivity and reduce errors. These cameras inspect products faster than humans and never tire. They spot tiny defects, collect real-time data, and help robots work safely. In automotive manufacturing, machine vision increases productivity by 21% and cuts scrap rates by 25%. Electronics factories see a 30% drop in missed defects and 40% shorter inspection times. Semiconductor plants reach 95% accuracy in defect detection. These improvements lead to higher quality products and lower costs.Image Source: statics.mylandingpages.coMachine vision systems operate around the clock, making manufacturing settings more efficient and reliable.Quality ControlEnhanced quality control stands as a key benefit of modern machine vision. Companies use these systems to check every product for defects or size errors. Machine vision software, valued at $1.76 billion in 2024 and projected to reach $2.1 billion in 2025, powers these inspections. Computer vision helps factories find flaws early, which reduces waste and saves money. In-process metrology uses non-contact gaging to measure parts within thousandths of an inch. This technology ensures that only top-quality items leave the production line.IndustryOperational Efficiency ImprovementDescriptionAutomotive Manufacturing21% productivity increaseBoost in production speed and outputAutomotive Manufacturing25% reduction in scrap ratesLess material waste and defective productsElectronics Manufacturing30% reduction in missed defectsImproved defect detection accuracyElectronics Manufacturing40% shorter inspection cycle timesFaster inspection processesSemiconductor Production95% accuracy in defect detectionHigh precision in identifying microscopic flawsEmerging UsesMachine vision now reaches beyond factories. In retail, AI cameras track inventory and enable cashierless checkout. Healthcare uses computer vision for medical image analysis, sometimes surpassing human accuracy. Doctors use these systems for early disease detection and surgery guidance. Smart cities rely on machine vision for real-time traffic analysis and public safety. Autonomous vehicles use computer vision to identify lanes, signs, and people. These new uses show how industrial image processing and computer vision shape daily life.The computer vision market is growing fast, with a projected 19.8% annual growth rate. This growth brings new solutions for many industries.Benefits and ChallengesAdvantages of Machine Vision CamerasMachine vision systems bring many advantages to modern industries. These systems help companies improve product quality and increase efficiency. For example, in agriculture, machine vision systems support precision farming by monitoring crops, guiding automated harvesters, and optimizing resources. Farms can reduce labor costs and improve crop health by using these cameras. In factories, machine vision systems allow for high-speed quality control and real-time data analysis. This leads to fewer mistakes and better products. Companies also use these systems for preventive maintenance, which helps avoid costly breakdowns. Safety improves because machine vision systems can monitor worker compliance and detect hazards quickly. New technologies, such as high-precision vision and 3D imaging, promise even more benefits in the future. These advancements will help industries reach higher levels of automation and sustainability.Machine vision systems help businesses save money, improve safety, and produce better products.Common ChallengesDespite many benefits, machine vision systems face several challenges. One major issue is meeting strict safety and reliability standards, especially in critical areas like self-driving cars or aircraft. Current image classification models have error rates around 1%, which is much higher than what safety standards require. For example, safety integrity levels (SIL) demand extremely low failure rates, as shown in the table below:Safety Integrity Level (SIL)Max Failure Rate (Low Demand)Max Failure Rate (High Demand)SIL 410^-4 failures per use10^-8 failures per hourSIL 310^-3 failures per use10^-7 failures per hourSIL 210^-2 failures per use10^-6 failures per hourSIL 110^-1 failures per use10^-5 failures per hourImage Source: statics.mylandingpages.coAnother challenge comes from dataset bias and testing methods. Many datasets do not represent real-world conditions, which makes it hard for vision systems to perform well in every situation. Testing and verifying deep learning models remains difficult because failure mechanisms are complex and not fully understood. Out-of-distribution detection and formal verification methods still cannot guarantee complete safety. These challenges show that machine vision systems must keep improving to meet the needs of high-risk environments.Machine vision cameras in 2025 help industries work faster and smarter. New technologies like AI, 3D vision, and edge computing make these systems more accurate and reliable. Market research shows strong growth, with a projected CAGR of 8.3% through 2028. Companies use these cameras for better quality control, less waste, and safer workplaces.AI and deep learning improve defect detection3D vision expands uses in robotics and healthcareEdge computing and 5G allow real-time decisionsMachine vision will shape the future of automation. Now is the time for businesses to explore these solutions.FAQWhat makes a machine vision camera different from a regular camera?A machine vision camera captures images for analysis by computers. It uses special sensors and fast processors. These cameras work in factories and labs. They help machines see details that people might miss.How do 3D machine vision cameras help in industry?3D machine vision cameras measure depth and shape. They help robots pick up objects and check product sizes. Factories use them for tasks that need precise measurements. These cameras improve accuracy and speed.Can machine vision cameras work in low light?Yes, many machine vision cameras use advanced sensors and special lighting. They capture clear images even in dim areas. Some models use infrared or event-based sensors for better results in low-light conditions.Are machine vision cameras hard to install?Most machine vision cameras come with easy setup guides. Many connect with USB or Ethernet cables. Companies design them for quick installation. Some systems use plug-and-play features for faster setup.What industries use machine vision cameras in 2025?Machine vision cameras appear in many fields. Factories, farms, hospitals, and smart cities use them. They help with inspection, safety, and automation. Retail stores and warehouses also rely on these cameras for tracking and quality control.

Electrical engineering and electronics traverse many fields of technological innovations and are in the foreground for groundbreaking advancements. Over the years, electrical and computer engineering have pioneered and contributed to developing more streamlined design, development, testing and improved manufacturing processes for frontier-end electronics, devices and equipment. In the strive for streamlining and in pursuit of innovation, the electrical and electronics industry has rapidly edged into the ever-expanding applications of artificial intelligence and machine learning.With the emergence of new technologies such as IoT, artificial intelligence (AI), machine learning (ML) and deep learning, the electronics and electrical industry is adopting and embracing major changes that lead to faster workflows, through optimization, automation and removing redundancies. ML and AI in industrial setups are designed to optimize systems and improve efficiency. This is possible as these systems are equipped with sensors and analytical processes that compute and interpret the data providing useful information.For instance, engineers create networks of interconnected cameras and sensors to guarantee that an autonomous vehicle's AI can "see" its surroundings. They must also make sure that the data is transmitted from these onboard sensors as quickly as possible because any lag in processing might cause a serious mishap.The electrical and electronic industry has seen significant advancements in recent years, with the emergence of artificial intelligence (AI) and machine learning (ML) revolutionizing the way we design, build, and operate electrical systems and devices. AI and ML technologies are increasingly being used to optimize performance, reduce costs, and improve efficiency across a range of applications in the industry. From predictive maintenance and fault detection to energy management and personalization, AI and ML are transforming the electrical and electronic industry in unprecedented ways. This article provides an overview of the role of AI and ML in the industry, examining their applications, benefits, and prospects.To start with let us understand what the terms AI and ML mean and how they are different.What are AI and ML?Artificial intelligence (AI) and Machine learning are often used synonymously although there are some differences between these two terms. To understand AI, it is important to first define what is machine learning and differentiate it from artificial intelligence.Artificial Intelligence is commonly known as AI, which refers to the development of computer systems, embedded systems and logical processing that can perform tasks that typically require human intelligence, such as learning, problem-solving, decision-making, and language processing.  For computer systems to learn and mimic human intelligence, they need to learn from a large dataset. For computers to learn from data, adapt to changing settings, and carry out activities that would otherwise require human involvement, artificial intelligence (AI) entails the creation of algorithms, machine learning models, and other approaches. Self-driving cars, voice recognition, image and speech recognition, fraud detection, and many more uses for AI are becoming more prevalent.Machine Learning or ML for short is a basic subset of AI. Basically, it involves the design and development of simple or complex algorithms and models that enable computers to learn from data, make analyses and improve their performance on a task without being explicitly programmed. Machine learning algorithms can learn from a large dataset to identify patterns, correlations and relationships, and then use that knowledge to make predictions or decisions on predictive new data. Machine learning can thus be described as a way of predicting the future based on presented parameters. There are three main types of machine learning: supervised learning, unsupervised learning, and reinforcement learning. ML is used in a wide range of applications, including image and speech recognition, natural language processing, recommendation systems, and predictive maintenance, among others.Artificial Intelligence and Machine Learning in the Electrical and Electronics IndustryThe term artificial intelligence as used in the electrical and electronics industry generally refers to a variety of systems and technologies built to imitate human intelligence by making decisions and solving related problems. In recent years, engineers and scientists have explored different applications and ways in which AI can be applied in electrical systems.Some of the most common ways in which AI is incorporated into electrical systems and consumer electronics:• Expert systems: Expert systems are a subset of artificial intelligence (AI) systems that employ an inference engine to derive conclusions from a knowledge base containing data on a particular area. If-then rules that are drawn from the knowledge of human subject-matter experts often make up the knowledge base. To offer a solution or suggestion, the inference engine applies these principles to the data or issue at hand. Since the 1970s, expert systems have been utilized for a variety of tasks, including financial planning, technical problem-solving, and medical diagnosis. They have the potential to successfully automate decision-making procedures and lessen the requirement for human expertise in specific jobs. They may not always deliver correct or timely information due to their limits in dealing with complicated or dynamic circumstances.• Fuzzy logic control: Fuzzy logic is a mathematical framework that deals with uncertainty and imprecision, and fuzzy logic control systems employ fuzzy logic to make choices and manage operations. Traditional control systems base judgments on exact numerical values, whereas fuzzy logic control systems base decisions on the degree to which linguistic variables are members of fuzzy sets. In complicated control systems, where it may be challenging to set precise numerical values for every input variable, fuzzy logic control systems are particularly helpful in cases when there is ambiguity or imprecision in the data.• Machine Learning: The creation of algorithms and models for machine learning (ML) enables computers to learn from data and enhance their performance on a job without being explicitly programmed. A huge dataset may be used by machine learning algorithms to detect patterns and correlations, which can subsequently be applied to fresh data to produce predictions or choices.• Artificial Neural Networks: A class of machine learning model known as artificial neural networks (ANNs) is modelled after the structure and operation of biological neural networks in the human brain. Artificial neurons, also known as neurons, are linked nodes that are arranged in layers to form ANNs. Each neuron in the network takes input signals from neighbouring neurons or outside sources, analyzes them using an activation function, and then generates an output signal that is sent to other neurons.• Deep Learning: To train artificial neural networks with many layers and enable them to learn hierarchical representations of the input data, deep learning is a subset of machine learning. Deep learning models are highly suited for a variety of applications, including image and audio recognition, natural language processing, and autonomous systems since they can automatically learn to discover complex patterns and correlations in the data. To alter the weights of the connections between the neurons in the network, deep learning models are often trained using a large dataset and an optimization technique, such as stochastic gradient descent.Application of AI and ML in the Electrical and Electronics IndustryThe design, construction, and use of electrical systems and devices are being transformed by artificial intelligence (AI) and machine learning (ML), which are becoming more and more relevant in the electrical and electronic industries. The industry is being impacted by AI and ML in the following ways:• Predictive maintenance: AI and ML algorithms may be used to identify when electrical equipment is most likely to fail and plan repair, cutting down on both maintenance costs and downtime.• Optimization: AI and ML models can be used to analyze large sets of data and make the decision based on the data. Machine learning algorithms can be used to optimize the performance of electrical systems and devices, ensuring that they are operating at peak efficiency. This optimizes processes, equipment and devices to perform more efficiently.• Fault detection and diagnosis: AI and ML algorithms can be used to detect and diagnose faults in electrical systems, allowing for more accurate and efficient troubleshooting and repair. The process can be based on the information collected by the sensor in systems or by predictive analysis based on previous parameters.• Energy management: Energy management is one of the most vastly used applications of AI due to its ability to compute and make a logical decision in homes and industrial setups. AI and ML can be used to optimize energy consumption in buildings and homes, reducing energy waste and saving money on energy bills.• Design optimization: AI and ML algorithms can be used to optimize the design of electrical systems and devices, improving performance and reducing costs.• Robotics and Automated Assembly: By automating the assembly of electrical components, AI and ML can increase productivity and decrease the need for manual labour. Algorithms can quickly and accurately detect and manipulate electrical components using computer vision and other approaches AI and ML are critical for developing and improving the performance of robots and automated systems used in manufacturing and other industries.• Personalization: Smart bulbs, smart homes and smart devices are based on the idea of customizing and personalizing technology. AI and ML can be used to create personalized electrical devices that adapt to individual user preferences, improving the user experience. These devices operate depending on the preferences of an individual or a particular setting thus making them customizable. • Smart Grid Management: The electricity grid can be managed more effectively with the help of AI and ML. Algorithms can find possibilities to decrease energy waste and suggest modifications to optimize the distribution of electricity by evaluating data on energy consumption trends.• Quality Control: Quality control in the traditional manufacturing process and assembly is a manual process that is often inaccurate and done through sampling. Throughout the manufacturing process, electrical component flaws may be automatically detected using AI and ML. Algorithms may find flaws and eliminate problematic parts by evaluating photos and other data, therefore raising the calibre of the final output.Overall, artificial intelligence (AI) and machine learning (ML) are revolutionizing the electrical and electronic sectors, allowing engineers to create, construct, and use electrical systems and gadgets more successfully and efficiently than ever before.FAQs1.What is the difference between AI and machine learning? AI refers to the broader field of creating intelligent machines that can perform tasks that typically require human intelligence, while machine learning is a subset of AI that involves training algorithms to make predictions or decisions based on data.2.How are AI and ML used in the electrical and electronics industry? AI and ML are used in a wide range of applications in the electrical and electronics industry, including predictive maintenance, process optimization, fault detection and diagnosis, energy management, and smart grid management.3.What are the benefits of using AI and ML in the electrical and electronics industry?The benefits of using AI and ML in the electrical and electronics industry include improved efficiency, increased productivity, reduced downtime, improved safety, and cost savings.4.What are the challenges of implementing AI and ML in the electrical and electronics industry? The challenges of implementing AI and ML in the electrical and electronics industry include the need for large amounts of high-quality data, the complexity of developing and training models, the cost of implementing new technologies, and the need for specialized skills and expertise.5.What are some popular AI and ML techniques used in the electrical and electronics industry? Some popular AI and ML techniques used in the electrical and electronics industry include artificial neural networks, fuzzy logic control systems, deep learning, reinforcement learning, and genetic algorithms.6.What are some examples of AI and ML applications in the electrical and electronics industry? Some examples of AI and ML applications in the electrical and electronics industry include energy demand forecasting, predictive maintenance of electrical equipment, automated fault detection and diagnosis, and optimization of power grids.

Warm hints: The word in this article is about 3000 words and reading time is about 15 minutesSummaryThis article is mainly about space monitoring and location technology of AR&VR equipment. All AR&VR products must have accurate spatial monitoring and positioning capabilities in order to achieve a full range of head， hand interactions in order to provide users with a more realistic virtual experience. CoreResearch on space monitoring and location technology of AR&VR equipmentPurposeIntroduce space monitoring and location technology of AR&VR equipmentProductAR&VR EquipmentCategoryElectronic productsApplicationsAR&VR technologyTechnologyInside-out space monitoring and positioning and outside-in space monitoring and positioning     Catalogs CatalogsI、Current Status of AR&VR ProductsV、 PC VRII、Cardboard productsVI、Space positioning technology categoryIII、Head motion controlled all-in-one            machine1. Inside-out space monitoring and positioning technologyIV、 AIO machine with space monitoring and positioning2. Outside-in space monitoring and positioning technology  Introduction I、Current Status of AR&VR ProductsThere are many AR&VR products already on the market. The price varies from Cardboards to Hololens, and the user experience varies greatly. First of all, VR products can be divided into the following categories: cardboard products, one-touch control integrated machine, AIO machine with space monitoring and positioning, PC VR.  DetailII、Cardboard productsCardboard productsThis kind of product is the simplest VR product. It comes from Google's carton VR box. Simply, it uses two convex lenses to project the contents of the mobile phone into screens. It uses the IMU inside the phone to judge the user's head movement to adjust the movement direction of the VR display content. This kind of product began to ship a large number of domestic white cards in 2015, but the overall effect of VR depends on the configuration of the mobile phone and the quality of the optical lens. It is only an entry-level VR experience product.The first generation VR products of game companies and some domestic video content are basically in this form. The most representative is the Storm Box, with a BT control handle for menu selection. The emergence of this type of product has indeed played a big role in the promotion of the entire VR industry, but its drawbacks are also very obvious. The Cardboard-style VR device ushered in a particularly big opportunity in 2016, that is, Google proposed Daydream VR. Google wants to enhance the VR Cardboard users’ experience, and hopes to be able to dominate the entire VR ecosystem in 2016. The Daydream mobile phone Pixel, Pixel XL and Daydream Viewer wearing the Cardboard were released in the year, and also equipped with a built-in IMU remote control handle. According to Google's requirements for VR device user experience, it must reach a mobile phone with a certain hardware configuration, and with Daydream Viewer running related VR applications, so that users will not be dizzy. Since 2016, with the launch of products such as Huawei P9 and Lenovo MotoZ, there are more and more mobile phones supporting Daydream. It is expected that, in 2017, these products will completely break the stalemate of the poor Cardboard user experience.III、Head motion controlled all-in-one machineAll-in-one machineAIO machine, as its name, is a VR device that does not need to be inserted into a mobile phone or connected to a PC. Currently, there are representative DaPeng, Xiaoniao, KanKan, etc. Most of the products are listed in 2016. The features of these products are basically the same use of 1080p or 2K resolution display screens (some of which also use 2Kx2 dual screens) with specially designed FIN products designed for VR products aspherical optical lens. The overall picture quality and distortion have improved a lot. However, these products have a common feature, that is, only a simple head control, can only sit or stand in place by shaking his head to achieve VR control, can not do the first hand interaction. In order to remedy this defect, there are currently some external accessories with spatial position detection (the built-in IMU handle, somatosensory backpack, spatial positioning light ball, etc.) to enhance the overall VR user experience.IV、 AIO machine with space monitoring and positioningAIO machine with space monitoring and positioningSpace monitoring and positioning are generally two ways, Inside-out and Outside-in. Inside-out senses the spatial information outside the device through the built-in sensor and performs related positioning. The Outside-in is just the opposite. It uses an external sensor to sense the position of the VR device and provides positioning information to the VR application. For the all-in-one machine, the current view has selected the Inside-out solution without exception, thus avoiding the need to install an external sensor in advance. The user can pick up the VR device at any time and anywhere. It is expected that some of the major products coming on the market in 2017, such as the second generation of the DaPeng AIO Machine and the Storm AIO Machine, will have space positioning capabilities, which will greatly enhance the VR user experience, especially VR games.V、 PC VRPC VRThe most representative products are the HTC Vive, Sony PSVR and HTC Oculus. After the three swordsmen went public in 2016, the HTC Vive has become a standard configuration for the VR experience room. Many other industry customizations are based on Vive. Sony PSVR has sold about 1 million units in less than half a year, which has greatly improved the overall level of user experience in the entire VR industry. The feature of these three products is the use of external space positioning technology. For example, Vive uses an external infrared laser transmitter, head and handle senses transmitter location to determine its position and trajectory. PSVR and Oculus use external sensors to sense the visible light or infrared light emitted by the headlights and the handle to determine the positions of the heads and the handles, so that the heads and the handles can accurately know their respective relative positions and trajectories in space at any time. This can make the application of VR more realistic.Next, let’s talk about AR products. At present, there is a big gap between AR products. Some AR glasses products only project virtual information, images, etc., and are not actually related to the actual environment where the user is located. Such products can only achieve 6DOF motion tracking by the projected virtual information following the head movement. But it is impossible to fuse the virtual object with the actual environment where the user is. For example, the famous ODG, the latest product only supports 6DOF motion tracking, there is no spatial positioning function.AR product ODGShanghai Zhishi's smart glasses also have similar functions to those of ODG, but their positioning is a live broadcast from the first point of view. Therefore, only relevant information and videos for shooting can be displayed in the glasses without spatial positioning. However, it is understood that they are also developing a built-in TOF depth camera glasses, which can achieve 6DOF + spatial positioning function.Shanghai Zhishi's smart glassesThe currently available AR products have the best space monitoring and positioning capabilities. The best match for virtual objects is the Lenovo Phab2 Pro based on Google's Tango space positioning technology and the Microsoft Hololens. Although these two products are very different in shape, they are essentially the full integration of 6DOF+SLAM, achieving spatial positioning and motion tracking. Their AR effects can basically be virtual objects and the actual environment. Convergence, no matter how the user moves or views from what angle, the feeling is just like the real object is in front of me. However, Hololens' accuracy and stability of 6DOF+SLAM are still higher than those of Tango in terms of the effect of actual use. This is closely related to the choice of hardware design platform, type and number of sensors, and core algorithms.Hololens ARIn addition, there are actually many companies in the country that are developing related AR products. Among them, the distinctive feature of Micro Eye's SMAKKEST is not only similar to Hololens, but also similar to Hololens in terms of function. The effect of 6DOF+SLAM is said to be recognized by Hololens.  AnalysisVI、Space positioning technology categorySpace monitoring and positioning technology, in fact, can be divided into two major categories initially, one is the Inside-out space monitoring and positioning, and the other is the Outside-in space monitoring and positioning.1. Inside-out space monitoring and positioning technologyInside-out space monitoring and positioning technology, technology is essentially similar to the process of human-to-environmentally-aware environment of the human eyewear. It senses the external space from the inside of the device and is therefore called the Inside-out positioning method. Inside-out space monitoring and positioning technology is also divided into two categories. One is the spatial monitoring and positioning based on SLAM technology. This is also a basic function required by the fiery robotics industry. SLAM is the English abbreviation for simultaneous localizatiion and mapping. It means that the device itself starts to move from an unknown location in an unknown environment, locates itself according to location estimates and maps during the movement, and builds incremental maps based on its own positioning to achieve autonomous detection and positioning navigation of the device itself. . At present, most AR products use Inside-out spatial location technology.The representative products of the space monitoring and positioning technology based on SLAM are Lenovo Phab2 Pro Tango mobile phone and Hololens. Both use depth monitoring + Fisheye and IMU fusion for sports monitoring. The difference is that the depth monitoring technology is different (Tango uses TOF, and Hololens uses structured light) and the number of Fisheye motion and feature point monitoring is different. Hololens' spatial positioning effect is more accurate and stable than Tango's, and Hololens' effect is in low light and glass light transmission. The current Tango victory, but Tango technology is not static, the next generation of Tango technology also supports multiple Fisheye camera.The Lenovo Phab2 Pro Tango phone uses a TOF depth camera, a 155-degree FOV Fisheye motion monitoring camera and an RGB camera, Fisheye motion monitoring camera with IMU data fusion, feature point matching, providing the device with a complete trajectory The monitoring, together with the cloud image information of the TOF depth camera, can plot the position of the device itself in space and the trajectory of the movement in real time. At the same time, through Tango's own zone learning function, when returning to the space that has already passed, once the Fisheye camera By detecting the stored feature points and matching them, it is possible to quickly perform spatial positioning. This is very important for AR equipment and it ensures that users can quickly use it anytime, anywhere. The working principle of Tango is not explained here. You can refer to Lin Shigong's another article for detailed understanding.The working principle of TangoThe Hololens design of Microsoft is very unique and very affluent. It has newly developed an HPU (Holographic Processing Unit) based on Intel's CPU and GPU technology. The device adopts the configuration of CPU+GPU+HPU. Like Tango, it does not need to External PCs look to achieve complete space monitoring and positioning. On the sensor side, depth monitoring uses two structured light cameras with four structured light-grating emitters for depth monitoring. In addition, there are two Fisheye cameras on the left and right of Hololens, so that more feature points in the space environment can be monitored. Hololens' space positioning accuracy and stability are therefore much higher than those of Tango.HPU（Holographic Processing Unit）The other type is based on the marker (pointer) space monitoring and positioning technology, simply put some like two-dimensional code, special color graphics patterns or light spots as markers in the actual space, equipment to monitor these markers The location of the point to determine their position and movement trends. The HTC Vive is a representative of this type. The following figure shows the working principle of spatial location of the HTC Vive.The working principle of spatial location of the HTC ViveVive emits laser light through two fixed laser emitters. Each base station has an infrared LED array. Two infrared laser emitters with rotating shafts perpendicular to each other are distributed. One is X-axis scanning, and the other is Y-axis scanning. The lasers have a fixed phase difference of 180 degrees. When one of the lasers emits an X-axis scan, the other laser emits a Y-axis scan.The HTC Vive has a shell that is densely covered with 32 light sensors, each oriented in different directions. These sensors are used to receive the infrared laser emitted by a fixed laser emitter. The computer connected to Vive controls all the devices to operate synchronously. The light sensor on the head reveals the time when the laser light reaches each sensor in the X-axis and Y-axis directions and the phase relationship between the two different laser emitters, and the phase difference of each light sensor can be calculated. Accurately locate the head position and trajectory.There are 24 light sensor on the Vive handle. The working principle and head of the handle are the same. It will not be repeated here.The HTC Vive ShellIn addition, it is necessary to mention that domestic Ximmerse, they also introduced a VR device for the Outside-in space positioning package design, through this space positioning package, can provide space-free VR devices with similar accuracy with the HTC Vive Space monitoring and positioning capabilities. The positioning package includes a dual camera module, two control handles with a light ball. When used, the dual camera module is externally mounted on the head display. The dual camera captures the light ball of the control handle to determine the position of the handle in the space, and the handle Through BT to transmit its 6DOF information to the module, the module fuses the 6DOF and the position information of the ball to achieve omni-directional positioning, thereby improving the interaction of the first hand. The advantage of this solution is that under any ambient lighting conditions, the positioning accuracy is not affected by the ambient light, and the positioning accuracy is high. However, this solution cannot monitor and locate the overall space environment in which the user is located. It can only monitor the spatial position information of the photo ball handle, but its advantage is that the original VR device without space positioning can have at least similar HTC Vive. Spatial positioning capabilities.Ximmerse2. Outside-in space monitoring and positioning technologyOutside-in space monitoring and positioning technology is currently the VR device is a relatively mature space positioning technology, Sony PSVR and Oculus are based on a similar program, but Oculus is in an external infrared camera plus point initiative marks the way to get Higher accuracy and faster response time. The PS VR directly uses the PS 3's MOVE system. The principle is similar to that of the Kinect. An external binocular depth camera is used for motion recognition and tracking. There is a light indication on the head, and a different color light ball is at the top of the handle. The external dual camera checks the trajectory of the headlight and the ball on the handle at any time. At the same time, the PS StaTIon receives the IMU information of the head display and handle returned by the BT. , Through the calculation you can get a complete headshot and handle movement trajectory and positioning in space.PS StaTIonFrom the principle and the effect of actual use, the space monitoring and positioning precision of HTC Vive is not high, although Sony's gaming and PSVR wearing wearing their own moderate than Vive better, but the spatial positioning accuracy is not high, it affects the VR The overall experience of the game. So we see from Sony's official website a new spatial orientation of Sony patent, this patent is to be ready for the next generation of PVSR, from the description of the patent point of view should be similar to HTC's fixed laser emitters way, with this program The current practice of PSVR is the opposite. Instead of using an external camera to take pictures (outside-in positioning), the external VR signal is used to locate the direction of the VR helmet in space. This can provide similar HTC Vive usage accuracy (probably higher), and can capture the user's various poses and trajectories 360 degrees without dead ends.Outside-In positioningOculus space monitoring and positioning technology with Sony PSVR similar difference is PSVR camera uses a dual monitor visible external program, and is an active Oculus uses infrared light to the head and the handle are placed significant infrared emitting LED , Through an external infrared camera for shooting and capturing the headlights and infrared light spot information on the handle, so as to obtain head trajectory and handle movement trajectory and spatial position information.Whether it is Sony PSRV or Oculus's Outside-in program, there is a problem, that is, the external camera is fixed, and when the user is back to the camera, the camera can not detect the head and the handle on the light, may There is a possibility of missed supervision, thus affecting the user experience, which does not exist on the HTC Vive.Another Marker point-based Outside-in space monitoring and positioning technology is Ximmerse's photosphere with dual-camera solution. This is contrary to their previously mentioned Inside-out solution, where dual camera monitors have been placed. Indoor fixed position, and then put the ball in the VR head and handle, head and handle during the movement of a dual camera monitor recorded head and handle in the space of the three-dimensional position and their movement data, through BT or WiFi is passed to VR heads to perform related operations and processing, and feedback is implemented on the screen to allow them to freely move around in the virtual environment and interact with things in the virtual world.Lightball + Dual Camera SolutionAnalog of Lightball + Dual Camera SolutionHere is a video about Augmented Reality vs Virtual Reality:Augmented Reality vs Virtual Reality - AR vs VR | The Future ! | Tamil Tech  Book RecommendationVirtual, Augmented, and Mixed Realities in EducationThis book describes the current state of the art of various types of immersive learning: in research, in practice, and in the marketplace. It discusses advanced approaches in the design and development for various forms of immersive learning environments, and also the emerging innovations in assessment and research in the field.--by Dejian Liu, Chris Dede, Ronghuai Huang, John Richards Experience on Demand: What Virtual Reality Is, How It Works, and What It Can DoVirtual reality is able to effectively blur the line between reality and illusion, pushing the limits of our imagination and granting us access to any experience imaginable. With well-crafted simulations, these experiences, which are so immersive that the brain believes they’re real, are already widely available with a VR headset and will only become more accessible and commonplace. But how does this new medium affect its users, and does it have a future beyond fantasy and escapism?--by Jeremy Bailenson  Relevant information about "Research on Space Monitoring and Location Technology of AR&VR Equipment"About the article " Research on Space Monitoring and Location Technology of AR&VR Equipment", If you have better ideas, don't hesitate to  write your thoughts in the following comment area. You also can find more articles about electronic semiconductor through Google search engine, or refer to the following related articles. 

SummarySingaporean researchers,led by by professor Hirotaka Sato,describe their work about designing robots--It's possible to use a living insect as a platform to develop a living insect-machine hybrid robot.Such a hybrid retains the complex structure of the insect's rigid exokeleton,complaint joints,and soft actuators, as well as the insect’s locomotion capability, and it does so while enabling high controllability and low power consumption. Such an insect-machine hybrid robot is made of a living insect platform with a miniaturized electronic device attached on it to control it.   By using the insect itself as the robot, researchers bypass the complex processes of designing and fabricating the robot body, using the insect’s muscular system as the soft actuators and flexible joints and its nervous system as part of the control system. About BeetleThis kind of particular beetle is a a darkling beetle. It’s small (2 to 2.5 centimeters), lightweight (about 0.5 gram), and lives for three months or so, which is a long time for a little bug. A backpack of electronics interfaces with the beetle’s antennae, and when the antennae are stimulated with an electric pulse, it activates the beetle’s built-in escape mechanism, fooling it into thinking it’s running into something and causing it to turn. The picture is from Nanyang Technological University AdvantageThe advantage of doing things this way (as opposed to direct nerve or muscle stimulation, something that the researchers also experimented with) is that the beetle’s brain is still in charge of controlling its limbs such that it’ll respond to high-level controls with adaptive gaits and such, making locomotion a much simpler problem to solve. With just two coin cell batteries, the cybeetle can be controlled for 8 hours, which is long enough for it to travel over a kilometer at an average speed of 4 cm/s. The following picture is from  Cyborg Insect: Ultralightweight Living Legged Robot The key to effectively controlling an insect using these methods is that the response to the antenna stimulation can’t be binary, since you’d end up with a level of control that would often be too coarse to be useful. By changing the frequency of the stimulation, the researchers were able to modulate how sharp of a turn the insect took: Increasing the stimulation frequency also increased the insect’s turning rate, with a success rate of over 85 percent. Stimulating both antennae at once causes the insect to back up, and it moves forward by default, giving you just about as much control as you can hope for. Living Robots' DifferencesElectrical stimulation is commonly used for neuromuscular stimulation in cyborg insects such as cockroaches, giant beetles, and moths. There are other groups working on antenna stimulation but they were not able to grade the response of the insect, which is very important for developing a precise closed-loop control system to make the cyborg insect work autonomously. The giant cyborg beetle mainly relies on neuromuscular stimulation of direct flight muscles for flight control and leg muscles of the fore legs for walking control. Ideally, stimulating the muscle would be more precise as we can perfectly control the individual legs, but it costs more in implantation and computing to plan and stimulate all the individual muscles for walking. Antenna stimulation is simpler and easier than stimulating all the individual muscles thus it helps us to simplify the hardware and control system a lot. Hopefully, in the near future, we can control the cyborg beetle as precisely as any other artificial motor. The zophobas beetle were used to develop this cyborg insect because its small size (2-2.5 cm) would help it to access the small rubbles system easily at disaster sites, where the cockroach and giant beetle can not get in. Moreover, a swarming of flying and walking cyborg insects of various sizes would increase the coverage and reduce the searching time, thus enhancing the efficiency and accuracy of search and rescue operations. Control IssueFor walking cyborg insects, researchers are able to integrate external sensors into the backpack as the insect is able to carry loads up to double its weight. We are developing a new backpack with integrated sensors for human detection and navigation. It would help us to detect victims when using cyborg insects at disaster sites, and enable the cyborg insects to work autonomously. On the other hand,research could release hundreds of flying and crawling cyborg insects to the sites as the price for one cyborg insect would be negligible once mass produced for a disaster scenario.The insects can move freely themselves into the collapsed structures and send back maps of their positions and environmental conditions so that the rescue team can plan for their action efficiently on how and where they should access. Once an insect detects a victim, it will send an alarm to the rescue team and switch to autonomous control mode to move around the victim for confirmation and build a clearer map of surrounding environment. At the end of the rescue operation, all the insects will autonomously return to the control base. I know that it sounds like science fiction, but we are in fact working to realize it. Researcher's GoalNow,researchers are working on a feedback control system to precisely control the insect locomotion with high reliability. We are also developing a new backpack with a navigation system and environmental sensors designed to promote fully autonomous and practical cyborg insects. For real applications, we need to maintain the power supply for the cyborg insect (mainly for the electronics backpack), which is currently a huge challenge if we just rely on the battery. So we are developing a biofuel cell, which is able to convert biofuel inside the insect to electric current for running the control backpack. It will help to maintain the backpack power for long-term use. Article resources: journal Soft RoboticsAtticle edited by kynix