現今圖形人機系統(GUI)都是架構在作業系統上。應用程式一般透過作業系統提供的繪圖函式庫繪製出使用者互動介面。此一方法通常需要大量的硬體計算資源,即使是一個簡單的應用功能,也需要借助完整的作業系統和函式庫的支援,不但耗費硬體成本,也增加系統耗電。本研究基於低階嵌入式系統使用的微控制器(MCU)為平台,開發一套無須作業系統的圖形使用者介面的嵌入式中介軟體。此一中介軟體包含兩個核心模組:圖形LCD顯示控制、以及基本的點、線、圓、矩形等基本繪圖模組及觸控手勢辨識模組。前者提供工業人機介面系統開發者藉以快速設計出豐富2D視覺化元件(visual gauge),後者則整合觸控面板控制和手勢辨識,但本專題最終目的是要設計出透過MIAT實驗室一套方法論[1][2]以實驗板提供給GUI基本的畫點、畫線、畫圓、畫矩形…等的架構,此方法可加速GUI在顯示時的速度也可大幅降低佔用硬體資源,且配合液晶顯示器模組(Liquid Crystal display Module,LCM)上的觸控面板來操控人機互動式介面,但最終的GUI介面可透過遠端Server編輯圖形佈署至為Client端的實驗板上力求方便、快速及美觀性。 Most of the GUI(Graphical User Interfaces) are being built on dedicated host operating systems, and applications generally rely on the graphics middleware of their operating systems to draw and map out user interfaces. This method usually requires a lot of computational hardware resources and RTOS-level supporting even for a simple GUI application, this is not only inefficient, but also increase the system power consumption.Our study aims to develop a HMI embedded middleware with no requirement of operation system, build on directly bare-metal microcontroller platform. This middleware consists of two core modules: GUI module and touch gesture module. In the GUI module, we implement the graphic LCD display control and fundamental 2D graphics functions, which provide the HMI developer to quickly design rich 2D visual gauges. For touch gesture module, we integrate the touch panel control and gesture recognition functions. The MIAT design methodology is applied to design the embedded HMI middleware. According the hierarchical and modular design paradigm, we implemented an efficient GUI and touch gesture API for Cortex-M3 microcontroller. A HMI prototype system integrating the proposed middleware is realized. It reveal that our system can accelerate substantially the GUI display and touch-based human-machine interaction and simultaneously reduce the hardware resources.