請選擇領域
請選擇部會
23 筆結果
技術簡介: • 技術創新：持有多模態人機互動技術專利，專業打造「能聽、能說、看得見」的 3D AI 虛擬人，並可與智販機、健身器材、血壓機、機械手臂以及 AR 眼鏡等設備串聯，提供豐富的人機互動服務，有效解決人力短缺問題並提升顧客體驗。 • 競爭優勢：與微軟、NVIDIA 等國際大廠建立正式合作，獲得技術與商務支持，輔以多模態AI人機互動技術專利護城河，逐步形成全球級的競爭優勢。 • 重要事蹟：已成功應用於「衛武營、Computex、禮客Outlet、桃園機場、大阪萬博 Taiwan Tech Week」等大型場域。 產業應用性: 本系統已廣泛應用於智慧零售、醫療、金融、文化展演等場域，支援多語系、年齡與性別等互動數據分析功能，具備高度可擴展性與應用彈性 ．零售百貨：智慧導購、商品推薦、智慧櫃台 ．醫療保健：虛擬諮詢、醫療掛號、衛教說明 ．金融服務：智能客服、業務諮詢、產品推廣 ．文化展演：展覽導覽、票務服務、展品解說 ．影視藝文：AI虛擬偶像、3D虛擬MV、AI節目主持人
未來科技館 | 生活應用技術簡介: 為了因應6G Massive MIMO通訊頻譜效率提升以及陣列天線數目增加的應用需求，工研院開發出創新的四極化共構天線專利技術。運用此天線技術在相同陣列面積下，能成功整合2倍天線數量，產生2倍資料波束。同時自主開發的6G FR3頻段之射頻前端晶片，具備高線性度、高效率等特性。透過晶片與天線模組緊密結合設計，更可達到降低連接損耗、提升模組效率的優勢。 產業應用性: 從5G進展到6G技術，射頻晶片需求預期將呈現數十倍的增長。射頻前端晶片成本將佔6G 基地台RU設備材料清單成本(Bill of Materials, BOM)的30%至60%。因此，掌握自主的射頻前端晶片技術，將大幅提升台灣基地台產業的競爭力。這項6G FR3射頻晶片組與天線陣列技術的先期開發，可協助國內晶片製程廠商、晶片廠商與天線模組廠等上下游產業鏈的串聯，共創6G商機。。
未來科技館 | 電子與光電技術簡介: 1. 賦予SSD生成式AI推論能力，兼顧SSD原有的儲存功能，透過更換SSD即可實現AI推論。 2. 分散主GPU算力需求，極大化AI算力的使用效能及效率。 3. 實現百工百業邊緣AI應用普及。 產業應用性: 目前全球生成式AI因建置費用昂貴及雲端資安問題，尚未真正進入百工百業，本計畫將研發全球第一個整合6nm PCIe 6.0 AI運算儲存控制晶片及AI推論服務器，未來透過置換SSD即可在地端實現生成式AI推論，可運行超過30B的大語言模型推論應用，以滿足未來AI應用遍地開花所需之地端推論需求。
未來科技館 | 電子與光電技術簡介: 數位發展部為建立符合國家與產業需求之人工智慧(AI)產品與系統評測機制，並健全國內AI評測制度之發展環境，於2023年12月成立AI產品與系統評測中心(AIEC)。AIEC以兩大主軸「評測環境」與「國際接軌」為核心工作，建立我國自主AI認驗證體系與符合國際規範的評測系統，提供產業AI驗測服務，進而接軌國際達到相互認可之目標。 產業應用性: 建立我國AI評測體系，結合國際標準(如NIST, ISO)與臺灣特色(繁體中文)，協助公私部門打造可信任AI服務。透過首座AI測試實驗室與驗證機構，建立國際互認基礎；並以標準化流程提供第三方測試報告，做為政府採購與企業導入依據。首波聚焦語言模型五大評測項目（公平性、準確性、可靠性、隱私、資安），後續將拓展至影像與多模態領域。
未來科技館 | 資訊與通訊Technology Introduction: AlixLabs' unique and patented Atomic Layer Pitch Splitting (APS) is a novel plasma-based etching method that selectively targets surface topology. This breakthrough technology eliminates several complex steps in chip manufacturing, enabling more efficient, cost-effective, and sustainable semiconductor production. Industry Applicability: APS enables cost-efficient scaling of semiconductor devices by doubling pattern density. It reduces manufacturing costs, lowers energy use, and supports sustainable, high-volume production. Applicable in logic, memory, and power devices, APS accelerates industrial adoption with scalable chip manufacturing.
未來科技館 | 材料化工與奈米Technology Introduction: Bertha is a 4nm AI prcocessor with LPDDR5x, designed for efficient Generative AI inference. Since infernece is bottlenecked by memory, Bertha has streamlined memory access and LLM-optimized compute dataflow to achieve 90% bandwidth utilization compared to GPU's 50%. It features highly scalable chip-to-chip networking to accelerate models of any scale. Bertha delivers 2x higher performance, 3x lower power consumption, and 10x better cost than GPUs. Industry Applicability: Bertha significantly reduces the total cost of ownership of cloud service providers when running Generative AI services and encourages wider adoption of AI in more diverse industries. Bertha is expected to replace the expensive GPUs from the datacenters for greater sustainability. Since Bertha is scalable, we also apply Bertha in edge computing to enable personalized AI in IoT devices and home appliances.
未來科技館 | 資訊與通訊技術簡介: Cupola360智慧巡檢解決方案結合4K/8K全景影像與AI邊緣運算，採用信驊科技專利全景拼接晶片打造智慧巡檢相機，提供具高度擴充彈性的AI 平台。透過 Cupola360⁺ 系統，可實現 360 度零死角巡檢，並支援火煙偵測、人流監控、群眾行為分析等 AI 應用。此方案能即時發現異常，協助企業提升安全與管理效率，廣泛應用於智慧工廠、智慧城市、智慧零售等場域，加速產業數位轉型。 產業應用性: Cupola360智慧巡檢解決方案如同Eyes of AI，透過360度全景影像與AI技術，可廣泛應用於智慧工廠、智慧城市、智慧零售等百工百業，能即時監測火煙、人流與行為異常，避免潛在風險並降低事故發生。管理人員可遠端即時掌握現場狀況，不需親臨現場即可完成巡檢，提升效率並節省人力成本。此方案不僅能加強安全防護，也能協助企業改善營運流程，加速數位轉型，適用於製造業、公共設施、交通場域及大型零售環境。
未來科技館 | 機械與系統Technology Introduction: PurCity's patented GapS facade panels revolutionize buildings. They act as substitutes for traditional exteriors, drawing in polluted air via natural ventilation, and purified via a harmless chemical process, trapping PM2.5 (+50%), PM10 (+90%) & VOCs. Integrated with HVAC, GapS provides cleaner, colder air (-5 to -10°C, proven at Toyota pilot), cutting HVAC energy use by up to 60%. It is integrated with AI/ML to ensure peak performance. Industry Applicability: PurCity's technology applies to various industries, including construction, HVAC, and manufacturing. It's implemented in new builds and retrofits to improve energy efficiency and air quality. Benefits include significant energy savings, reduced carbon footprint for ESG compliance, and enhanced brand value through verifiable sustainability. It's a key solution for companies facing carbon taxes and seeking to create smarter, healthier buildings.
未來科技館 | 綠能與環境技術簡介: •展示Genraemon系統影片生成操作，輸入Prompt後一鍵生成。 • 展示結合台灣在地元素的影片生成，並可同步展示加入音效與口白，調整音量並整合影片內容。 • 進行生成式AI基礎模型之相關侵權風險識別與案例蒐整，並初步建構平台侵權風控機制與技術研發邏輯。 產業應用性: 現有雲端通用型生成式 AI 服務缺乏在地化特色，且在腳本撰寫、影片剪輯與音效整合等應用上仍有瓶頸。透過 Genraemon 系統進行在地化影片生成，協助國內各類規模業者降低行銷成本、提升品牌國際競爭力，同時推動國內 AI 技術發展，帶動新興數位內容服務與相關就業機會。
未來科技館 | 資訊與通訊技術簡介: - <0.5s Instant Response — Ultra-low latency interaction '- On-Premise AI — Works without internet connection, ensuring data privacy '- Multimodal Interaction — Speech-to-Speech and Vision integration '- Complete Edge AI Toolchain — Developers can easily create tailored experiences by our labeling and finetuning tools to train and build domain-specific knowledge '- Built-in Echo and Noise Reduction IP 產業應用性: DeepMentor's latest innovations are transforming human-machine interactions by moving away from button-based interfaces to seamless, natural conversations. This voice-first approach provides a hands-free, intuitive, and accessible UI/UX for people of all ages and abilities. It is expected to be integrated into applications such as smart cockpits, smart home appliances, companion robots, and interactive kiosks.
未來科技館 | 機械與系統Technology Introduction: 3D Architech’s core technology is a gel-based metal 3D printing process that achieves 10-micron resolution—10× finer than conventional methods—enabling custom, high-performance cooling and energy components at a fraction of the cost. Industry Applicability: Our gel-based metal manufacturing technology enables micro-architected copper cold plates tailored for AI chip cooling. It achieves custom micro-architected design with 10-micron precision, reduces electricity costs for data center cooling by up to 60%, and supports rapid customization for new chip designs, driving efficiency and scalability in the AI industry.
未來科技館 | 材料化工與奈米技術簡介: EndoSemio's AI-powered ENT endoscope fuses narrow-band and white light imaging to highlight vascular changes linked to early nasopharyngeal cancer. Using a CNN+LSTM AI model trained on a large clinical database, it enables rapid, real-time identification of high-risk lesions—bringing advanced cancer screening to ENT clinics and reducing diagnostic delays. 產業應用性: This technology enables ENT clinics to adopt AI-driven cancer screening without relying on expensive imaging infrastructure. By integrating real-time vascular analysis into standard ENT endoscopy, it lowers diagnostic barriers in high-risk regions. It supports early detection, reduces specialist workload, and accelerates industry adoption of intelligent diagnostic tools.
未來科技館 | 生技與醫療Technology Introduction: We combine artificial intelligence with sensor hardware to enable smarter, more context-aware systems. These technologies are revolutionizing sectors like autonomous vehicles, drones, robotics, healthcare, security, and consumer electronics by turning raw sensor data into actionable intelligence. Industry Applicability: AI-driven sensing refers to the integration of machine learning or deep learning algorithms into systems that collect data from the physical world—using sensors and RF transceivers.
未來科技館 | 電子與光電技術簡介: - Power-efficient, high performance RF chips on silicon CMOS including beamformer (BF), power amplifier (PA), low-noise amplifier (LNA), frequency converter and controller chip for AESA system '- Separate sets for different frequencies: Ku-band, Ka-band, and higher frequency bands '- Modular CMOS RF blocks allow mix-and-match, reuse, and integration (BF+PA, LNA+PA, etc.) 產業應用性: AESA controls phase and amplitude of individual component in an antenna phase array electronically, eliminating need for mechanical parts and extremely fast response time (<100nS). Beamformers and RF front end ICs are the most important components in an AESA. AESA can be used in LEO communication system (UT & satellite), and also for radars.
未來科技館 | 資訊與通訊Technology Introduction: Polaris Electro-Optics is pioneering high-speed silicon-organic modulators to overcome data rate bottlenecks in conventional silicon photonics. Our proprietary Ferroelectric Nematic Glass (FNG) material exploits the Pockels effect and unique silicon photonic modulator design to enable fast, energy-efficient optical modulation. Industry Applicability: Our technology is developed to address the surging demand for high-speed data transfer across data centers, telecommunications, AI infrastructure, and optical computing. FNG is integrated onto silicon photonic ICs post-foundry, without modifying standard fabrication processes, delivering superior performance, manufacturability, and scalability.
未來科技館 | 電子與光電Technology Introduction: femtoAI's core technology is an Embedded AI Platform that uses sparsity to achieve ultra-efficient neural network inference—skipping unnecessary computations and storage. This results in up to 10x space and 100x energy efficiency gains. Our full-stack solution spans custom AI accelerator silicon and production-ready algorithms, co-designed to bring real-time intelligence to everyday devices. Industry Applicability: femtoAI’s ultra-efficient Embedded AI Platform powers intelligence in devices across industries—from audio and wearables to smart appliances. Our custom accelerator, SPU-001, together with AI models for speech enhancement and voice commands, delivers real-time, low-power intelligence that makes products smarter, smaller, and more sustainable.
未來科技館 | 電子與光電Blumind's proprietary all-analog semiconductor architecture mimics the human brain and delivers the most efficient power profile possible for AI inferencing applications. Blumind’s IP implements a relative ratiometric quantized approach to critical AI operations i.e MAC, activation, decision and propagation operations, completely in the analog domain and with a single transistor! Blumind's IP solves process, voltage temperature and drift challenges, inherent in analog AI solutions, while maximizing performance and reliability. Blumind can reduce power consumption for most inferencing applications by 100-1000+ times compared with today’s digital chips. Blumind has received several global awards and recognitions for its disruptive approach to AI processing. Blumind was the recipient of the 2023 best startup at Computex Taiwan, MWC top 50 award 2024 Spain, EEtimes 100 top startups 2024 USA, SWITCH top 10 startup 2024 Singapore, Techradar top 10 AI hardware companies 2025 USA.
未來科技館 | 電子與光電技術簡介: 為了因應先進封裝所帶出的各種新興設計問題，本計畫提出一連串的EDA工具鍊，從異質整合系統的整合及效能評估，到全晶片的熱模擬技術及功能時序分析，一直到晶片、封裝與PCB的實體設計與優化，我們都有對應的規劃，希望可以解決此類晶片的設計瓶頸。最後，我們也會跟工研院合作，將本計畫所產出的成果串接到工研院開發的EDA平台之中，形成一個完整的EDA流程供廠商使用，以真正發揮協助國內產業開發先進封裝產品的目標。 產業應用性: 先進封裝技術被認為是進一步提升晶片計算能力的關鍵，但目前的EDA工具多由國際大廠主導研發，缺乏自主性，難以有效協助本土公司解決先進封裝所帶來的各種新興設計問題。因此，我們試著結合學界的EDA研發能量，提前佈局下世代之新興晶片設計軟體技術，透過開發一系列的工具鏈，為國內產業建立先進封裝相關的EDA技術及設計流程，並透過工研院的open EDA平台，協助企業導入這些自動化技術，成功開發先進封裝之產品。
未來科技館 | 電子與光電敬請期待!
