Category: Artificial intelligence

  • Xiaomi to Integrate CyberOne Humanoid Robot in Manufacturing

    Xiaomi to Integrate CyberOne Humanoid Robot in Manufacturing

    Xiaomi launched its first humanoid robot, the CyberOne, in 2022. This robot features high power density actuators and is part of Xiaomi’s Cyber series, which also includes the CyberDog and CyberDog 2.

    Xiaomi to Integrate CyberOne Humanoid Robot in Manufacturing

    Integration into Manufacturing

    Recently, Xiaomi Robotics has been actively working on incorporating the robot into its manufacturing system. The company believes that the intelligence of these robots can be used in specific manufacturing scenarios. Xiaomi envisions creating an open-source platform through technological advancements and breakthroughs.

    Future Developments

    The division is focused on further developing the technology. A more advanced and intelligent version of the robot could contribute to various aspects of Xiaomi’s manufacturing process. Xiaomi also plans to extend the robot’s applications to industries such as electronics, machinery, artificial intelligence, and automobiles. This project is expected to become a significant demonstration of humanoid product development, manufacturing, and application in Beijing.

    CyberOne’s Capabilities

    The CyberOne robot is equipped with sophisticated arms and legs, supports bipedal motion, and can reach a peak torque of up to 300Nm. It also has the ability to detect human emotions and boasts advanced vision capabilities, allowing it to create a three-dimensional virtual reconstruction of the real world. The robot stands 177cm tall, weighs 52kg, and has an arm span of 168cm. It supports up to 21 degrees of freedom in motion, with a real-time response speed of 0.5ms for each degree, enabling it to accurately mimic human movements. The self-developed humanoid bipedal control algorithm ensures that CyberOne’s walking posture is natural and stable.

    “With AI at its core and a full-size humanoid frame as its vessel, this is an exploration of possibilities of Xiaomi’s future technological ecosystem and a new breakthrough for the company,” said Lei Jun, Founder, Chairman, and CEO of Xiaomi Group.

    Tags: , ,
  • OPPO to Add 100+ AI Features to Smartphones by Year’s End

    OPPO to Add 100+ AI Features to Smartphones by Year’s End

    Oppo is making a significant investment in artificial intelligence (AI) with the intention of making advanced smartphone features more accessible to millions of users. The company has announced its plan to incorporate over 100 generative AI features across its entire range of smartphones by the end of 2024, targeting an estimated 50 million users.

    This ambitious initiative is supported by the creation of a dedicated AI R&D Center in Shenzhen. This center will specialize in key AI areas such as image processing, computer vision, and natural language processing. Oppo has also affirmed its commitment to user privacy, assuring that user data will not be used for training models and that strong data security measures will be in place.

    Strategic Partnerships

    Oppo's AI strategy is heavily reliant on strategic partnerships with Google, Microsoft, and MediaTek. Google’s Gemini family of large language models (LLMs) will be incorporated into Oppo’s flagship smartphones, facilitating features like AI Writer and AI Recording Summary. AI Writer will provide content suggestions, grammar checks, and multi-language translation, while AI Recording Summary will summarize meetings and podcasts, emphasizing key points.

    Microsoft will contribute its expertise in voice and text conversion. Oppo users will gain advantages from Microsoft Fast Transcription for seamless audio recording transcription and enhanced voice assistant interactions using Neural TTS. Furthermore, Oppo and Microsoft are collaborating on Desktop Copilot, enabling users to create content, translate text, and search for information on their smartphones through a connected PC.

    Hardware Optimization

    The collaboration with MediaTek is focused on hardware optimization. The development of chipsets with dedicated AI processing units (NPUs) will ensure smooth performance and efficient AI processing. Custom AI frameworks will also be developed to utilize MediaTek’s capabilities while enabling distinctive AI features on Oppo devices.

    Oppo is confident that the shift towards AI-integrated phones is a long-term trend that will revolutionize the mobile experience. The company envisions a future with intelligent operating systems equipped with AI agents and multimodal interaction, paving the way for a fully developed AI phone ecosystem.

    Tags: , ,
  • TECNO Showcases AIoT Innovations at COMPUTEX 2024

    TECNO Showcases AIoT Innovations at COMPUTEX 2024

    TECNO has made a significant impact at COMPUTEX 2024 by unveiling a variety of AI-enhanced products under the theme “Smart of Creation.” The standout devices in TECNO’s presentation include the Dynamic 1 robotic dog, PocketGo, Mini PCs, MEGABOOK Series, and the AI-powered CAMON 30 Premier 5G.

    TECNO AI and AR Innovations

    Among the notable products is the TECNO PocketGo, an AR PC handheld that recently secured the Red Dot 2024 Product Design Award. This device merges AR glasses with a handheld PC, transforming mobile gaming. With the AMD® Ryzen™ 7 8840HS processor and advanced head tracking, it offers a 6D immersive gaming experience. The PocketGo integrates smoothly with Windows and connects with TECNO smartphones and mini PCs.

    Dynamic 1 Robotic Dog

    Another highlight from TECNO is the Dynamic 1 robotic dog. This bio-inspired quadruped, which resembles a German Shepherd, comes with AI interaction capabilities. Equipped with an Intel® RealSense™ D430 depth camera and multiple sensors, it excels in obstacle detection and navigation. The Dynamic 1 can climb stairs, jump, and respond to voice commands, exemplifying TECNO’s innovation in robotics.

    CAMON 30 Premier 5G

    The CAMON 30 Premier 5G enhances video calling and imaging with AI features like AIGC Portrait, AI Erase, and Universal Tone. The new CAMON 30 Premier LOEWE Edition, introduced at the exhibition, features an eco-friendly design with a coffee grounds back cover and green color accents.

    Mini PCs and MEGABOOK Series

    TECNO showcases the world’s smallest water-cooled gaming MINI PC G1, which includes an Intel® Core™ Ultra processor, 32GB DDR5 RAM, and 1TB SSD for superior performance. The MEGA MINI M1, a compact desktop weighing just 445g, delivers up to 45W performance.

    The MEGABOOK T16 Pro 2024 Ultra, TECNO’s first AI PC, features an Intel® Core™ Ultra 7 Processor, a 99Wh battery, up to 32GB LPDDR5 RAM, and 1TB storage. It offers AI functionalities such as AI Image Generation and AI Copilot.

    Mega Interconnect, Smart Manager

    TECNO’s PC Manager software improves collaboration between TECNO laptops and smartphones, streamlining workflows and optimizing device performance. The OneLeap feature allows seamless data sharing, multi-screen collaboration, and file management, ensuring data security and efficiency.

    TECNO’s varied range of innovative products at COMPUTEX 2024 highlights its dedication to pioneering the future of technology with AI and AR advancements.

    TECNO Showcases AIoT Innovations at COMPUTEX 2024
    TECNO Showcases AIoT Innovations at COMPUTEX 2024
    Tags: , ,
  • Chinese Scientists Create World’s First Multifunctional Electronic Skin

    Chinese Scientists Create World’s First Multifunctional Electronic Skin

    Scientists at Tsinghua University in China have reached a significant milestone in artificial skin technology by developing the world’s first “electronic skin” with a bionic three-dimensional structure. This breakthrough seeks to replicate the extraordinary ability of human skin to sense touch.

    Mimicking the Spatial Distribution of Mechanoreceptors

    Human skin contains a complex network of specialized cells known as mechanoreceptors, which are arranged in three dimensions. These cells enable us to differentiate between various sensations such as pressure, friction, and strain with high accuracy. Existing electronic skin technologies have not been able to replicate this intricate and simultaneous perception at a physical level.

    Inspired by the natural design, Professor Zhang Yihui’s research team at Tsinghua University has created a new electronic skin model featuring a three-dimensional structure. This design emulates the spatial distribution of mechanoreceptors like Merkel cells and Ruffini corpuscles found in human skin. Consequently, the device can independently measure pressure, friction, and strain, representing a significant advancement in artificial touch technology.

    Three Distinct Layers

    This innovative electronic skin is carefully designed with three distinct layers that mirror the structure of human skin:

    • Epidermis: The outermost layer, similar to the surface texture of human skin.
    • Dermis: The middle layer, housing most sensors and circuits. It features a unique eight-arm cage structure that holds the force-sensing unit. This highly sensitive unit, positioned near the surface, effectively detects external forces.
    • Subcutaneous Tissue: The bottom layer, mimicking the texture of the underlying tissue in human skin.

    By integrating this three-dimensional electronic skin with advanced deep machine learning algorithms, the research team has accomplished something extraordinary. Their creation can not only sense touch but also analyze an object’s modulus (stiffness) and local principal curvature (shape) through simple contact.

    Potential Applications

    The groundbreaking research, titled “A Three-Dimensionally Architected Electronic Skin Mimicking Human Mechanosensation,” was published in the prestigious scientific journal Science on May 30, 2024. This development holds tremendous potential for a variety of applications, such as prosthetics that provide a more natural sense of touch for amputees and robots with improved tactile abilities.

    Tags: , ,
  • AI Avatars to Attend Meetings Soon, Says Zoom Founder

    AI Avatars to Attend Meetings Soon, Says Zoom Founder

    In a recent conversation with The Verge, Eric Yuan, the founder of Zoom, shared his forward-looking ideas for the future of video conferencing, including an innovative concept – digital twins for meetings.

    Eric Yuan envisions Zoom going up against Microsoft and Google soon

    Previously an employee at Cisco, Yuan founded Zoom with the intent of streamlining video conferencing. Now, he aims for Zoom to expand its horizons beyond video calls. Yuan envisions Zoom competing with tech giants such as Microsoft and Google in enterprise software, providing a full suite that includes document collaboration, email services, and chat functionality in addition to video conferencing.

    Investment in AI

    The interview also highlighted Zoom’s substantial investment in Artificial Intelligence (AI), a trend that is rapidly gaining traction across numerous industries. Yuan believes AI has tremendous potential in the workplace, particularly through the creation of “digital twins.” These AI-driven avatars could represent users in Zoom meetings, potentially even making decisions on their behalf while the user handles more urgent tasks.

    User Control Over Digital Twins

    Yuan, however, stresses the importance of users maintaining control over these digital twins. He imagines a future where users can train their digital twins to display different traits depending on the meeting context. For instance, a user might have one digital twin designed for sales presentations and another for technical discussions. While he acknowledges the uncertainties that come with AI development, Yuan is optimistic about the eventual realization of this technology.

    It is important to note that Zoom has already addressed user concerns about AI development in its terms of service.

    Tags: , ,
  • Arm to Deploy 100 Billion AI Devices by 2025

    Arm to Deploy 100 Billion AI Devices by 2025

    Arm Holdings is working towards deploying 100 billion Arm devices by 2025. This is part of a larger plan to secure a significant position in the fast-growing AI chip market, which is expected to expand from $30 billion in 2024 to more than $200 billion by 2032.

    Establishing a Dedicated AI Chip Division

    To reach this goal, Arm will create a dedicated AI chip division. This new division plans to produce prototypes by spring 2025 and start mass production by fall of the same year. Manufacturing will be handled by contract manufacturers, potentially including Taiwan Semiconductor Manufacturing Corp (TSMC) among others. The development costs for these AI chips are anticipated to be in the hundreds of billions of yen, supported financially by Arm and SoftBank.

    Strategically, there is a potential for the AI chip business to be spun off and integrated under SoftBank once mass production begins. This aligns with SoftBank CEO Masayoshi Son's vision of transforming SoftBank into an AI powerhouse. Son has allocated $64 billion to drive innovation in sectors such as data centers, robotics, and power generation. He aims to integrate AI, semiconductor, and robotics technologies to revolutionize industries like shipping, pharmaceuticals, finance, manufacturing, and logistics.

    New AI Chip Designs and Collaborations

    At the Computex forum in Taipei, Arm reiterated its goal of achieving 100 billion Arm devices globally by the end of 2025, focusing on AI-ready devices. Arm also introduced new AI chip designs and software tools intended to enhance AI capabilities in smartphones. These new designs are expected to boost compute and graphics performance by over 30% and speed up AI inference by 59% for machine learning and computer vision tasks.

    Arm’s strategy now includes providing ready-for-manufacturing blueprints, a significant change from their previous approach of offering abstract designs. This shift is designed to speed up the development process for partners like Samsung and TSMC. Samsung is already collaborating with Arm on 3nm process technology to meet the increasing demand for generative AI in mobile devices. TSMC is working with Arm to improve AI performance and efficiency through their Open Innovation Platform ecosystem.

    Enabling Customers

    Through these initiatives, Arm aims to position itself not as a competitor but as an enabler, assisting its customers in bringing AI-driven chips and devices to market more quickly. Chris Bergey, Arm’s General Manager, outlined the company's vision of combining a platform for tightly coupled accelerators with customer Neural Processing Units (NPUs) to accelerate the development of powerful AI chips and devices.

    As Arm progresses with its plans, it aims to capture a significant portion of the burgeoning AI chip market, filling the gap currently dominated by Nvidia. The company’s focus on AI chip development is a crucial component of SoftBank’s broader strategy to lead the AI revolution and transform various industries through advanced technology.

    Tags: , ,
  • iOS 18: Easily Remove Unwanted Objects from Photos Before Sharing

    iOS 18: Easily Remove Unwanted Objects from Photos Before Sharing

    Apple may soon introduce on-device and cloud-based AI features to iOS 18 for the iPhone 16 series, following in the footsteps of Google, Samsung, Huawei, and other Android OEMs. The heart of Apple's AI strategy for iOS 18, known as Project Greymatter, will integrate these features across core apps and services like Safari, Notes, and Photos.

    New AI Features in iOS 18

    One of the standout features is 'Greymatter Catch Up,' a notification summarization tool that allows users to listen to summaries of recent notifications via Siri. Additionally, Apple plans to introduce AI-powered photo editing and other features.

    iOS 18 Generative Playground

    Apple is reportedly working on a new object-erasing photo editing feature. This generative AI tool can recognize and remove unwanted objects from images, filling the space with AI-generated content that blends seamlessly with the surroundings. This feature is similar to what is already available on Pixel, Galaxy, and Huawei devices.

    Apple has also developed an "internal application" called Generative Playground, which may allow users to experience this AI-powered photo editing feature. This tool might be integrated into iMessage for easy sharing. iOS 18 could also offer the capability to generate personalized emojis.

    The new operating system will include an automated response generation feature based on large language models (LLMs). This feature will intelligently consider factors like names, calendar events, locations, and times when generating responses. Siri is also being improved to offer more natural and smooth text-to-speech functionality for a better user experience.

    Major Improvements to the Notes App

    The Notes app will receive a significant upgrade with iOS 18 and macOS 15, incorporating text-based generative AI capabilities like speech-to-text conversion and summarization. Users will be able to record audio directly from the Notes app, convert it to text, and summarize it. This could make it easier to access information from an entire meeting or lecture.

    Apple is expected to announce these developments at the upcoming WWDC conference, which will take place from June 10 to June 14.

    Tags: ,
  • Chinese Researchers Create Chip Mimicking Human Visual System for AI

    Chinese Researchers Create Chip Mimicking Human Visual System for AI

    Chinese researchers have recently made a notable advancement in brain-inspired chip technology with the introduction of Tianmoc. This innovative chip, developed by Tsinghua University, is designed to replicate the human visual system, offering enhanced performance in intricate environments.

    Tackling Visual Perception Challenges

    Traditional chips often face difficulties with abrupt light changes or interference, which can impede the functionality of self-driving cars and other technologies. Tianmoc addresses this challenge effectively by providing high-speed, high-precision processing while maintaining low power consumption, making it ideal for such demanding situations.

    Inspired by the human brain, Tianmoc decomposes visual data into “primitives” that are then reassembled, closely imitating human perception. This “dual-pathway” method outperforms conventional approaches, ensuring the system’s stability.

    Broad Applications and Future Potential

    Beyond its applications in self-driving cars and robots, Tianmoc holds the promise of transforming everyday devices like smartphones by introducing new capabilities as production costs become more affordable.

    Building on Previous Success

    This breakthrough builds upon the team’s earlier success with Tianjic, another brain-inspired chip. The researchers foresee Tianmoc and its predecessors contributing to a comprehensive “brain-like intelligence ecosystem,” which could be a cornerstone for developing artificial general intelligence—an AI that learns and reasons like humans.

    In summary, Tianmoc represents a major milestone in the advancement of brain-inspired AI technology, with the potential to address complex problems and lead to innovative applications.

    Tags: , ,
  • Huawei Challenges Nvidia in China’s AI Chip Market

    Huawei Challenges Nvidia in China’s AI Chip Market

    The competitive environment in China’s AI chip market is experiencing a major transformation as Huawei begins to challenge Nvidia’s dominance. This shift is largely driven by US sanctions on Huawei and restrictions on Nvidia’s advanced processors, which have reshaped the market dynamics.

    Huawei’s internally developed 910B Ascend chips are quickly gaining popularity across multiple sectors, thanks to the rising use of “AI boxes.” These AI boxes are integrated units that combine AI chips, industry-specific algorithms, and pre-trained large AI models, offering ready-to-use solutions for smaller businesses and government agencies. They provide a cost-effective option for these entities to fine-tune services based on large models without hefty investments.

    Huawei's Market Expansion

    Previously, Nvidia dominated the mainland market, but now there's a potential shift with Huawei likely to surpass its American competitor. In reaction to US export restrictions on its A800, H800, and L40S processors, Nvidia has launched lower-grade chips like the H20 specifically for China. These chips are sold at a discount compared to Huawei’s Ascend 910B chips, highlighting the increased competition and market challenges Nvidia is facing.

    Huawei’s Ascend ecosystem has expanded rapidly, with 40 hardware partners, 1,600 software partners, and 2,900 AI application solutions. Major Chinese firms like Tianjin Sitonholy and SenseTime are adopting Huawei chips for their AI boxes. SenseTime even expects a doubling of its generative AI sales due to these AI boxes.

    Analysts' Predictions

    Experts predict that Huawei’s growing AI chip business will soon dominate the market. AI boxes, considered crucial for businesses, are anticipated to feature primarily local players led by Huawei. Insights from Pacific Securities and Counterpoint analysts emphasize Huawei’s technological prowess and significant market potential. However, the speed of Huawei’s rise depends on its production capabilities, as manufacturing high-performance AI accelerators and server processors is more complex than mobile system-on-a-chip products.

    Despite these advancements, Nvidia remains a major player, having captured 90% of AI chips sold in China during the first half of the previous year, compared to Huawei’s 6%. Nonetheless, the market is shifting, with more Chinese companies launching AI boxes using Huawei chips. Nvidia’s tailored-for-China chips, including the H20, are currently sold at a discount compared to Huawei’s offerings.

    Future Outlook

    Moving forward, Huawei is expected to focus more on AI chips rather than mobile chips to secure a leading position in China. Collaborations with firms like iFlyTek aim to tackle computing-power limitations, with products like the Spark Number 1 AI box set against Nvidia’s A100 processor.

    Nvidia’s strategic response to this competitive pressure includes price reductions and the introduction of custom chips for the Chinese market. Despite these measures, analysts point out the uncertainty surrounding Nvidia’s future in China amid US-China tensions and a highly competitive market.

    As China’s share of the global AI industry is projected to exceed 30% by 2035, the ongoing rivalry between Huawei and Nvidia in the AI chip market will have significant implications for both companies and the broader industry.

    Tags: , ,
  • Qualcomm Unveils High NPU Scores for Snapdragon X Elite

    Qualcomm Unveils High NPU Scores for Snapdragon X Elite

    Qualcomm recently revealed remarkable benchmark results on May 28th, highlighting the AI capabilities of their Snapdragon X Elite processor’s Hexagon NPU. The Snapdragon X Elite X1E-80-100's NPU achieved a score of 1787 in the UL Procyon AI test, far outstripping its competitors: the M3 (898 points) and Ultra 7 155H (480 points).

    Significant Efficiency Gains

    Qualcomm asserts that the Hexagon NPU operates with much lower power consumption than its rivals. Specifically, it shows a 22% reduction compared to the M3 and a 31% enhancement over the Ultra 7 155H.

    Qualcomm has awarded the Snapdragon X Elite the title of “highest NPU performance per watt,” indicating an energy efficiency that is 2.6 times better than the M3 and an astonishing 5.4 times better than the Ultra 7 processor. Although these tests were carried out on Qualcomm reference designs and particular laptops, the results underscore the X Elite’s potential for efficient AI processing.

    Real-World Application Potential

    Qualcomm also demonstrated the GIMP + Stable Diffusion 1.5 raw image processing capabilities of the Snapdragon X Elite. In this scenario, the NPU provided 3 times the speed compared to the Core Ultra 7 155H, suggesting its promise for practical applications.

    Tags: , ,