➀ The F-Tac robotic hand achieves 100μm touch resolution across 70% of its surface using internal cameras and elastomer deformation detection, enabling precise object interaction;

➀ Its modular design combines 17 camera-equipped boxes with spring-loaded joints and wire-driven servos, allowing adaptive grasping through tactile feedback and machine learning algorithms;

➂ Developed jointly by Queen Mary University and Chinese institutes, the system demonstrated superior performance in 600 real-world trials compared to non-tactile alternatives.

➀ Empa researchers developed machine learning (ML) algorithms to optimize laser-based metal processing, such as 3D printing (Powder Bed Fusion) and welding, reducing costly preliminary trials by two-thirds through real-time optical data analysis;

➁ The ML system uses FPGA chips for precise, real-time control of laser parameters during welding, addressing material defects and variability to improve consistency;

➂ This approach enhances accessibility for non-experts, potentially enabling broader industrial adoption of advanced laser manufacturing techniques.

The ETH Zurich researchers have developed a method that makes AI answers more reliable over time. Their algorithm is highly selective in choosing data. Additionally, up to 40 times smaller AI models can achieve the same output performance as the best large AI models.

ChatGPT and similar tools often amaze us with the accuracy of their answers, but also often lead to doubt. One of the big challenges of powerful AI response machines is that they serve us with perfect answers and obvious nonsense with the same ease. One of the major challenges is how the underlying large language models (LLMs) of AI deal with uncertainty. It has been very difficult until now to judge whether LLMs focused on text processing and generation generate their answers on a solid foundation of data or whether they are on uncertain ground.

Researchers from the Institute for Machine Learning at the Department of Computer Science at ETH Zurich have now developed a method to specifically reduce the uncertainty of AI. 'Our algorithm can specifically enrich the general language model of AI with additional data from the relevant thematic area of the question. In combination with the specific question, we can then specifically retrieve those relationships from the depths of the model and from the enrichment data that are likely to generate a correct answer,' explains Jonas Hübotter from the Learning & Adaptive Systems Group, who developed the new method as part of his PhD studies.

Chinese researchers have built a 32-bit RISC-V processor using molybdenum disulfide (MoS2) on a sapphire substrate. The processor, RV32-WUJI, has 6000 transistors and operates at KHz speeds, executing the full RISC-V 32-bit instruction set. The researchers used machine learning to optimize the wiring and materials for the transistors. The overall yield was over 99.9 percent, with a chip-level yield of 99.8 percent.

➀ This article explores the implementation of an IoT-based health monitoring system using the statistical programming language R. The system aims to process and analyze real-time health data from various IoT devices, with a focus on predicting heart attack risks.

➁ The proposed system integrates wearable sensors, mobile applications, and cloud-based infrastructure to connect individuals with healthcare providers. R's analytical capabilities, including statistical functions, visualization tools, and machine learning models like GLMs, random forests, and decision trees, are utilized to generate insights and predictive models.

➂ The system emphasizes data security and privacy, incorporating encryption and authentication mechanisms. It also includes a user-friendly mobile app for real-time health monitoring and personalized feedback.

➀ This article introduces an AI-based predictive maintenance system designed to detect engine faults in vehicles through sound analysis. The system uses Google Teachable Machine to train an AI model capable of distinguishing between normal and abnormal engine sounds.

➁ The AI model runs in a browser, continuously monitoring the engine's sound. Upon detecting an anomaly, it triggers an immediate alert and sends an HTTP request to the IndusBoard Coin, which acts as a web server to process the alert.

➂ The IndusBoard Coin also hosts a webpage that visually indicates the system's status. If a fault is detected, the webpage turns red and displays an alert message while activating a blinking LED as a warning indicator.

Retrieval Augmented Generation (RAG) is making it easier to find information in extensive documents by using large language models (LLMs) and a retrieval system. This method ensures precise and comprehensive information retrieval, which is particularly useful for legal texts and user manuals. The Fraunhofer IWU is developing this technology, which can be used on standard PCs and in the cloud, ensuring data security and privacy.

➀ Mach42的发现平台利用机器学习创建一个代理模型，以实现更快的电路设计探索，而无需进行完整的SPICE模拟。

➁ 平台的目标是达到90%的准确率，允许快速迭代，同时保留在最终确认时进行完整准确性的选项。

➂ Mach42正在与Cadence合作开发Spectre，并计划开发Verilog-A模型，这可能会显著增强模拟-数字设计验证。

➀ The book, 'Statistical Machine Learning for Engineering with Applications,' is published by Prof. Dr. Anita Schöbel and Prof. Dr. Jürgen Franke.

➁ It aims to provide an accessible introduction to Machine Learning concepts and methods.

➂ The book includes detailed case studies from various industrial projects, focusing on practical applications and interpretation of Machine Learning methods.

➀ The book 'Statistical Machine Learning for Engineering with Applications' is published by Prof. Dr. Anita Schöbel and Prof. Dr. Jürgen Franke, providing an accessible introduction to the concepts and methods of machine learning.

➁ The book aims to familiarize readers with basic topics like classification trees, Bayesian learning, neural networks, and deep learning, emphasizing practical applications and interpretation over mathematical details.

➂ It includes several detailed case studies based on real industrial projects, covering a wide range of technical applications from vehicle manufacturing to process and material technology, and process optimization through image analysis.

➀ Aggressive testing in chip manufacturing often leads to the discarding of marginally functional chips, causing waste.

➁ Traditional testing methods like PAT have limitations in detecting subtle defects.

➂ proteanTecs' outlier detection solution uses machine learning to enhance chip reliability and performance.

➀ The reliability of AI-based image recognition is ensured through a smart validation environment, even under adverse conditions such as strong vibrations.

➁ Components and sub-systems can be efficiently characterized under various mechanical boundary conditions, for example, to easily generate AI training data.

➂ The transfer of developed technologies into the economy is demonstrated through various development steps along the Technology Readiness Levels (TRL).

➀ A research team in Saarland, Germany is developing miniaturized pumps and valves using dielectric elastomeric silicone films, which are lightweight, compact, and energy-efficient.

➁ These devices operate without compressed air, motors, or lubricants and are suitable for cleanroom environments.

➂ The technology is scalable and can be used in various applications, including automotive, medical, pharmaceutical, and industrial processing.

➀ Texas Instruments (TI) 发布了一款基于单芯片60GHz汽车雷达传感器AWRL6432的紧凑型低功耗参考设计，适用于车内应用。

➁ 这款高性能雷达系统可以实现儿童存在检测、入侵者检测、占用感应、驾驶员生命体征监测和安全带提醒等功能。

➂ 该设计具有小尺寸和优化的射频性能，为下一代车内传感应用提供高效解决方案。设计包括两个发射天线和三个接收天线，支持多种通信接口。

➀ Researchers at the Fritz-Haber Institute have developed the Automatic Process Explorer (APE), an approach that enhances our understanding of atomic and molecular processes.

➁ APE reveals unexpected complexities in the oxidation of palladium (Pd) surfaces, providing new insights into catalyst behavior.

➂ By using machine-learned interatomic potentials (MLIPs), APE predicts atomic interactions and improves the accuracy of simulations.

➀ An innovative full-body scanner linked to an AI diagnostic platform is being used to improve early detection of melanoma;

➁ The scanner, developed by project partner Bosch, scans the entire body in six minutes and uses cognitive AI to assess risks;

➂ Explainable AI (XAI) technologies are being incorporated to ensure the reliability and safety of the AI system.

The Fraunhofer-Gesellschaft has developed a new KI-Diagnosis Platform that improves the early detection of skin cancer using a whole-body scanner. The scanner, connected to the KI platform, analyzes the entire body in six minutes and provides a risk assessment for any suspicious skin changes. The project, iToBoS, involves 20 partners and is aimed at enhancing and speeding up the existing method of skin examination. The scanner uses cognitive AI to assist in the examination and provides a personalized risk evaluation for each mole. The platform integrates health data from various sources, and the project is supported by the EU with 12.1 million euros.

➀ Researchers at the University of Illinois Urbana-Champaign have developed a machine-learning framework called HUMANUP that enables humanoid robots to stand up autonomously after falling.

➁ The framework uses reinforcement learning (RL) to identify effective limb trajectories and refine initial motions into smooth and controlled movements.

➂ Tests in simulations and real-world settings using the Unitree G1 humanoid robot showed promising results, indicating the robot can autonomously recover from falls regardless of its position or the surface beneath it.

➀ The Ernst-Abbe-Hochschule Jena (EAH Jena) is establishing an endowed professorship for 'Explainable Artificial Intelligence (XAI)';

➁ The professorship aims to improve the transparency and comprehensibility of decisions made by AI systems through innovative methods and tools;

➂ The professorship will focus on interdisciplinary research and education, involving various fields such as healthcare, social services, and business administration.