| Abstract: | 步?導航系統(Pedestrian Navigation System) 其目的在於幫助使用者全球定位，指引使用者到達目的地，而經由導航系統定位及導航能夠?快速的規劃出?程?徑。近??得導航系統已成為人們生活中重要的一部分。一般導航系統主要是?用全球衛星定位系統(Global Positioning System, GPS)?進?定位，可顯示所在位置及?進的速?及方向。 GPS現今用於?輛導航上已相當成熟，但對於個人步?導航而言，?用導航系統除???於人們攜帶且GPS用於步?導航仍有限制與障礙，?如靜止或低速航?時，GPS 測得航向會有誤差，方向角會有偏差以及飄移之?況發生，無法?確用於個人徒步?進間的導航。為?克服GPS的限制，可藉由感測元件?輔助GPS的定位，如電子?盤(Electronic Compass) 可?測?進間的航向角?，而加速儀(Accelerometer) 可以偵測?進步伐的頻?及加速??進?導航。而步?導航問題的癥結就在步伐頻?及步距很難準確測?，每個人?走的自然頻?及步伐習慣皆?相同，而且也會隨?面的?同而改變，因此可以用人工智慧演算法?學習個人?走步伐頻?及步距之間的關係，讓導航系統有學習的能?，導入個人步?導航的應用。本計劃主要是結合GPS微機電系統與人工智慧等成為一步?導航系統。我們將應用蜜蜂演算法(Bees algorithm, BA)於個人步?導航系統中，當GPS 訊號?好時以GPS 訊號為學習對象，實作蜜蜂演算法進?學習；當GPS 訊號微弱?足以定位時，我們將蜜蜂演算法所學習的??進?方位推估演算，將GPS 之應用擴展至智慧型個人步?導航系統以準確定位。本研究以蜜蜂演算法學習個人步伐頻?及步距，配合實際GPS訊號做為修正以達到步?導航系統持續學習適應個人特質使之定位?為準確。而且針對個人步?時遭遇之?面情況智慧地進?個人步伐及步距估測做修正，使得系統在個人?走於?同?面如沙地、坡地時也能夠持續提供正確的導航定位資訊。
A Pedestrian Navigation System (PNS) is used to position a location and guide its user to a specified destination. The most widely used navigation system is the Global Positioning System (GPS), which provides the position, speed, and heading of the user. It is used in the military, aircraft and car navigation, and in land surveying, etc. However, the GPS cannot be used as a pedestrian navigation system since it is unable to precisely position in some circumstances, such as static heading, low-speed cruise, indoor walking, and so on. In order to overcome this deficiency in a pedestrian navigation system, some additional instruments and sensors are required to complement the GPS receiver. Inertial sensing devices such as gyroscope, magnetometer, and electronic compass are used to detect the direction of movement while accelerometer measures velocity and pedometer decides the number of steps moved and barometer measures the change in altitude. Thus, the traveled distance and direction can be estimated. Usually the direction can be easily identified; therefore, the precision of a traveled trajectory depends on the accuracy of stride estimation. The step length, which is affected by many factors, varies from person to person. It can be determined by applying artificial intelligence (AI) to determine the relationship between the step frequency and stride of a walker. That is, an intelligent pedestrian navigation system is capable of learning, and thus, is able to correctly calculate the traveled distance. The plan mainly combined with GPS, inertial sensing devices and AI, to become a PNS. The bees algorithm, which has the advantages of simplicity and stability, will be used to determine the step length in the system. Our method will be able to learn using GPS signals to obtain the parameter of a pedestrian by positioning. If the signals are weak or absent, the parameter will be applied to process dead reckoning. It will be focused on the calibration of the pedestrian navigation system, as well as increasing the precision of said system. The signal detected by the accelerometer is affected by varies types of movement, such as walking, running, and going up or down. In addition, it is also influenced by different types of ground over which the person moves. This system will be calibrated to be suitable for varies types of movement and ground. |
