aIwalk 智意行科技

智意行科技股份有限公司 · aIwalk Co., Ltd. aIwalk Co., Ltd. · 智意行科技股份有限公司
AI 驅動的數位孿生技術,打造從跌倒預防到完整行動支援的個人化復健生態系。
「透過生理數據,我們將克隆一個真正的你,並放到穿戴式機器人中,而達到 Lifelong Wearable Robot Partner」。
AI-driven digital twin technology for personalized rehabilitation — from fall prevention to full mobility support.
"Through physiological data, we will clone a real you and put it into a wearable robot, achieving the Lifelong Wearable Robot Partner."
3 核心產品Core Products
5 技術階段Tech Phases
11 團隊成員Team Members
2026 成立於Founded

關於智意行科技 About aIwalk Co., Ltd.

智意行科技透過三大核心產品開發 AI 驅動的復健科技:NTKCAP,一套可攜式無標記動作捕捉系統;ExoPulse Patient Cloud,臨床數據平台;以及 ExoPulse Exoskeleton,數位孿生驅動的穿戴式機器人。 aIwalk Co., Ltd. develops AI-driven rehabilitation technology through three core products: NTKCAP, a portable markerless motion capture system; ExoPulse Patient Cloud, a clinical data platform; and ExoPulse Exoskeleton, a digital-twin-powered wearable robot.
Ecosystem Overview Ecosystem Overview
(a) 個人化資料蒐集Personalized Data Collection
(b) 數位孿生資料合成Digital Twin Data Synthesis
(c) 部署與執行Deployment
Personalized Data Collection
Digital Twin Data Synthesis
Deployment
Close-Loop
ExoPulse media coverage preview
VERIFIED
系統可信度 System Credibility

不只自己說,台灣官方媒體實際報導。 Not just our word — covered by Taiwan's official media.

Overview detail

產品展示 Product Demos

NTKCAP
可攜式無標記動作捕捉系統Portable Markerless Motion Capture
ExoPulse Local APP
本地端應用程式Local Application

技術藍圖 Roadmap
核心願景:以 數位孿生 驅動的個人化外骨骼控制框架,從 IMU 感測到 數位永生,實現終身照護閉迴路 Core Vision: A personalized exoskeleton control framework driven by Digital Twin technology, from IMU sensing to Digital Immortality, achieving a lifelong care closed loop
我們的競爭優勢Our Competitive Edge
ExoPulse 在意圖預測與個體適應性雙維度上領先現有技術See how ExoPulse leads in intent prediction and individual adaptability
拖曳輔助線drag to read
1
IMU 感測
跌倒預防
IMU Sensing
Fall Prevention
Inertial Sensing
& Pre-Fail Detection
YEAR 1
衰弱前期個體
獨居長者
Pre-frail individuals
Elderly living alone
IMU Edge AI PCB FPGA IRB
2
EMG 驅動
偏癱復健
EMG-Driven
Hemiplegia Rehab
EMG-Driven NMS
Stroke Rehabilitation
YEAR 2
中風後偏癱
多發性硬化症
Post-stroke hemiplegia
Multiple Sclerosis
sEMG NMS Hill-type DRL 力矩倍率Torque Ratio
3
資料生成
族群擴展
Data Synthesis
Population Expansion
Data Synthesis
& Generalization
YEAR 3
重度失能族群
凍結步態 · TBI
Severe disability
Freezing of Gait · TBI
資料合成Data Synth NMS虛擬環境NMS Virtual Env Sim2Real 泛化Generalize
4
腦機介面
全癱瘓控制
Brain-Computer
Interface Control
BCI Integration
Full-Paralysis Control
YEAR 4+
完全癱瘓族群
脊髓損傷
Full paralysis
Spinal cord injury
EEG BCI Avatar 意圖鏈Intent Chain
5
數位永生Digital Immortality
Digital Immortality
Consciousness Continuity
LONG-TERM VISION
所有人
終身機器人夥伴
Everyone
Lifelong robot partner
數位分身Digital Twin 運動記憶Motor Memory AI延續AI Continuity SaaS
左右滑動翻閱卡片 Swipe to flip cards
Phase 1
IMU 感測與跌倒預防IMU Sensing & Fall Prevention
Inertial Sensing & Pre-Fail Detection
聚焦於 ExoPulse 系統之基礎感測架構建立。完成 PCB 佈局與原型機製作,整合 IMU 慣性感測模組與 Edge AI 即時運算晶片,實現步態相位即時估計與跌倒風險預警功能。此階段建立 ExoPulse 硬體平台基線,取得醫院 IRB 倫理審查核准,啟動不穩定步態族群之臨床資料蒐集。產品定位面向衰弱前期個體與獨居長者的跌倒預防,研究產出將直接轉化為企業之復健訓練與跌倒預防軟體模組。 Focuses on establishing the foundational sensing architecture for the ExoPulse system. Complete PCB layout and prototype fabrication, integrating IMU inertial sensing modules with Edge AI real-time computing chips to achieve real-time gait phase estimation and fall risk early warning. This phase establishes the ExoPulse hardware platform baseline, obtains hospital IRB ethics approval, and initiates clinical data collection for unstable gait populations. Product positioning targets fall prevention for pre-frail individuals and elderly living alone, with research outputs directly converted into rehabilitation training and fall prevention software modules.
階段產出PHASE OUTPUT
  • 復健訓練與跌倒預防軟體模組(Rehab Training & Falling Prevention)Rehab Training & Fall Prevention Software Module
  • ExoPulse 硬體平台原型機(U 型框架、三自由度髖關節、BLDC 馬達)ExoPulse hardware platform prototype (U-frame, 3-DOF hip joint, BLDC motors)
  • IMU 跌倒偵測與風險預警演算法IMU fall detection & risk early warning algorithm
  • 臨床 IRB 核准與基線資料蒐集系統Clinical IRB approval & baseline data collection system
目標族群TARGET POPULATION
衰弱前期(Pre-frail)個體、獨居長者 ─ 跌倒預防與安全保障Pre-frail individuals, elderly living alone ─ fall prevention & safety assurance
Phase 2
EMG 驅動的神經肌肉骨骼建模EMG-Driven Neuromusculoskeletal Modeling
EMG-Driven NMS Modeling & Stroke Rehabilitation
導入 sEMG 肌電感測與神經肌肉骨骼(NMS)模型校準流程,透過雲端平台完成個人化數位孿生建模。針對中風後偏癱患者之不對稱步態特徵,開發 EMG 驅動的即時肌力估測與輔助力矩計算演算法,完成軌跡補償連續調節之可行性驗證。DRL 雙代理架構中,Human Agent(TunedMPO)作為意圖預測引擎,Exoskeleton Agent(D4PG-LSTM)學習最佳人機互動策略。此階段產品定位由預防轉向主動復健與行動輔助 Introduces sEMG electromyographic sensing and Neuromusculoskeletal (NMS) model calibration workflow, completing personalized digital twin modeling via cloud platform. Targeting asymmetric gait characteristics of post-stroke hemiplegia patients, develops EMG-driven real-time muscle force estimation and assistive torque calculation algorithms, verifying feasibility of continuous trajectory compensation. In the DRL dual-agent architecture, the Human Agent (TunedMPO) serves as the intent prediction engine, while the Exoskeleton Agent (D4PG-LSTM) learns optimal human-robot interaction strategies. This phase shifts product positioning from prevention to active rehabilitation & mobility assistance.
階段產出PHASE OUTPUT
  • 監測與關節保護模組(Monitoring & Joint Protection)Monitoring & Joint Protection Module
  • 行動輔助軟體模組(Mobility Assistance)Mobility Assistance Software Module
  • EMG 驅動即時肌力估測系統EMG-driven real-time muscle force estimation system
  • DRL Human Agent + Exoskeleton Agent 雙代理架構DRL Human Agent + Exoskeleton Agent dual-agent architecture
  • MS 族群臨床資料庫建立MS population clinical database establishment
目標族群TARGET POPULATION
中風後偏癱患者、多發性硬化症(MS)族群 ─ 步態復健與肌力輔助Post-stroke hemiplegia patients, Multiple Sclerosis (MS) population ─ gait rehabilitation & muscle strength assistance
Phase 3
資料生成技術與族群擴展Data Synthesis & Population Expansion
Data Synthesis & Cross-Population Generalization
聚焦於數位孿生之前向資料合成技術(Forward Data Synthesis),利用已校準之 NMS 模型在虛擬環境中合成使用者未被實際記錄的運動場景數據(如不同地形、不同退化程度之步態模式),解決無法行動族群的臨床資料稀缺問題。使 ExoPulse 系統可服務於無法進行大量實體測試的重度失能族群——包含凍結步態(Freezing of Gait)患者及創傷性腦損傷(TBI)患者。驗證架構泛化與批量個人化能力,實現多地形、多疾病條件之 Sim-to-Real 轉移 Focuses on Forward Data Synthesis technology for digital twins, using calibrated NMS models to synthesize movement scenario data in virtual environments that were never physically recorded (e.g., gait patterns across different terrains and degradation levels), solving the clinical data scarcity problem for immobile populations. This enables the ExoPulse system to serve severely disabled populations who cannot undergo extensive physical testing ── including Freezing of Gait (FoG) and Traumatic Brain Injury (TBI) patients. Validates architecture generalization and batch personalization capabilities, achieving multi-terrain, multi-condition Sim-to-Real transfer.
階段產出PHASE OUTPUT
  • 資料合成引擎 ─ 虛擬場景運動資料批量生成系統Data Synthesis Engine ─ batch motion data generation in virtual scenarios
  • 跨族群泛化模型 ─ 一套框架批量應用於不同疾病族群Cross-population generalization model ─ one framework for batch application across disease groups
  • 凍結步態(Freezing of Gait)控制模組Freezing of Gait (FoG) control module
  • Sim-to-Real 驗證報告與臨床轉化數據Sim-to-Real validation report & clinical translation data
目標族群TARGET POPULATION
無法自主行動之重度失能族群、凍結步態(FoG)患者、創傷性腦損傷(TBI)患者Severely disabled populations unable to move independently, Freezing of Gait (FoG) patients, Traumatic Brain Injury (TBI) patients
Phase 4
腦機介面整合與全癱瘓族群控制BCI Integration & Full-Paralysis Control
BCI Integration & Full-Paralysis Exoskeleton Control
此階段為長期願景之探索性研究,探討結合腦電訊號作為運動意圖之最上游輸入源。使完全喪失自主運動能力之癱瘓患者可透過腦機介面(BCI)控制數位孿生虛擬人體,進而驅動外骨骼執行目標動作,形成 EEG → 虛擬人體意圖預測 → 外骨骼力矩輸出之完整控制鏈。同時探索癱瘓者透過意念控制虛擬人物(Avatar)的可能性,將 ExoPulse 服務對象進一步擴展至完全癱瘓族群,實現「Train. Assist. Live.」。 This phase represents exploratory research toward the long-term vision, investigating the integration of brain electrical signals as the most upstream source of motor intent. Enabling fully paralyzed patients who have lost all voluntary motor ability to control a digital twin virtual body through Brain-Computer Interface (BCI), which then drives the exoskeleton to execute target movements, forming the complete control chain: EEG → Virtual body intent prediction → Exoskeleton torque output. Also explores the possibility of paralyzed individuals controlling virtual avatars through thought, further extending ExoPulse services to fully paralyzed populations, realizing “Train. Assist. Live.”.
階段產出PHASE OUTPUT
  • EEG → 外骨骼完整控制鏈原型系統EEG → Exoskeleton complete control chain prototype system
  • 虛擬人體(Avatar)意念控制模組Virtual body (Avatar) mind control module
  • 下一世代產品線技術基礎 ─「Train. Assist. Live.」Next-generation product line technology foundation ── “Train. Assist. Live.”
  • BCI 數位孿生整合驗證報告BCI digital twin integration validation report
目標族群TARGET POPULATION
完全癱瘓族群、脊髓損傷患者 ─ 透過意念重獲行動自由Fully paralyzed populations, spinal cord injury patients ── regaining mobility through thought
Phase 5
數位永生Digital Immortality
Digital Immortality — Consciousness Continuity
ExoPulse 從 Phase 1 到 Phase 4 累積的個人化數位孿生——涵蓋神經肌肉骨骼模型、運動意圖模式、腦電訊號特徵、肌肉激活模式、步態習慣——構成了一個人最完整的身體運動數位分身。當這些數據與 AI 模型持續學習並演化,使用者的運動特徵、反應模式、甚至意圖預測都能在數位世界中被忠實重現與延續。這不僅是復健或輔助,而是邁向人類運動記憶的永久保存——一個真正的「Lifelong Wearable Robot Partner」的終極意義:Train. Assist. Live. Forever. The personalized digital twin accumulated by ExoPulse from Phase 1 to Phase 4 ── encompassing neuromusculoskeletal models, motor intent patterns, EEG signal features, muscle activation patterns, and gait habits ── constitutes a person's most complete physical movement digital avatar. As these data and AI models continuously learn and evolve, the user's movement characteristics, response patterns, and even intent predictions can be faithfully reproduced and continued in the digital world. This is not merely rehabilitation or assistance, but a step toward permanent preservation of human motor memory ── the ultimate meaning of a true “Lifelong Wearable Robot Partner”: Train. Assist. Live. Forever.
階段產出PHASE OUTPUT
  • 個人運動數位分身 ─ 持續演化的終身 NMS + 腦電模型Personal movement digital avatar ── continuously evolving lifelong NMS + EEG model
  • ExoPulse Lifelong Archive ─ 運動記憶雲端永久保存系統ExoPulse Lifelong Archive ── cloud-based permanent motor memory preservation system
  • 數位分身跨載體遷移技術(機器人、虛擬世界、下一代外骨骼)Digital avatar cross-platform migration technology (robots, virtual worlds, next-gen exoskeletons)
  • SaaS 終身照護訂閱服務SaaS lifelong care subscription service
目標族群TARGET POPULATION
所有人 ─ 從亞健康到完全癱瘓,每個人都值得擁有一個理解自己的終身機器人夥伴Everyone ── from sub-health to full paralysis, every person deserves a lifelong robot partner that understands them