全方位移動平台網路化控制及感測融合實現

Abstract

網路化在工業上的應用越來越普及，由於網路化可以帶來需許多優點，比如減少配線、減少安裝及維護的成本，所以由網路連結與多系統的溝通也成為了一種趨勢。在建構網路化控制方面，本研究運用RS-232串接的形式將多軸馬達連接構成一個網路化的系統，藉由DSP F2812與馬達之間的通訊協定成功的來控制四軸馬達。使平台行駛於低速(30 cm/s)與高速(100 cm/s)時所產生的實際位置誤差，並加入平台位置迴路上加上PID控制器，改善實際位置誤差。由於網路式的控制受限於傳輸率的限制，所需的取樣周期較大，因此我們將針對於不同的取樣時間下，探討網路式與與集中式的效果。 近來全方位攝影機在機器人導航領域中被拿來做為目標物追蹤以及機器人定位，本文以Kalman filter針對影像目標物相對位置作估測，以消除平台行走間產生震盪問題，本論文提出以感測融合的方式，結合影像與encoder，提升平台相對定位的精確度及穩定性。 本論文整合遠端控制、影像傳輸與平台運動控制三方面，遠端遙控者可以藉由網路傳輸下達命令給平台，平台上影像也可藉由網路傳輸回傳至遙控者觀看。最後，整體平台藉由網路化的結構，並具由遠端遙控及自主的系統，於全方位移動平台上實現。

Recently, network control system (NCS) becomes popular in different industrial applications with advantages as low cost, reduction of installation, and easy maintenance. The servo motor adopted in this study with the serious RS-232 interface to connect all motors as the NCS. The DSP F2812 is applied to generate commands as the protocol to control four motors of an omni-directional mobile robot (ODMR). The mobile robot was tested with both at a low speed (30 cm/s) and a high speed (100 cm/s) with acceleration and deceleration. By adding the PID position controller, the robot motion accuracy is significantly improved in a centralized control structure which is usually faster compared with the NCS structure. A suitable sampling time for the NCS on the ODMR is determined in real application experimentally by considering the sampling time and the system performance. The omni-directional camera can be applied as the function of tracking and localization on the robot navigation. Its estimation of relative position between an object and a robot is improved by applying the Kalman filter to reduce the oscillation as robot starts to move. Moreover, the sensor fusion is applied to integrate both the wheel encoder and the image signals to improve the accuracy and stability of localization. The present integrated system, including the remote control, the image, and the motion control systems, is developed for the remote operator to transmit the command and receive the image through the internet. Finally, the mobile robot which is built by the NCS structure has been applied to an autonomous system successfully.