自主式投餌機器人定位與導航系統之研發

Abstract

本研究為研發陸上養殖場用之自主式水上投餌機器人，整合定位和導航控制技術達成完全地自主性，以改變傳統陸上養殖場均以人工經營的模式並提升管理效率。投餌在養殖場是一個非常重要的過程，投餌時間應依照生物的特性，自主式投餌機器人能於白天或夜間以繞行養殖池四周的方式自動投餌，由機器人來減少人力的支出和負擔。 陸上養殖場的養殖池通常是以四面圍牆構成，為了讓機器人自主性的在養殖池上導航和定位，機器人必須能夠感知養殖池的環境和自身所處的位置。導航的基本任務就是投餌機器人沿著牆壁繞行養殖池一周，導航系統可以精確的測量投餌機器人的方位角和與牆壁保持的距離，超音波具有自動追蹤牆壁功能，實驗結果顯示機器人的與牆壁夾角大於±45度時均能有效的偵測到與牆壁的距離。距離和方位角作為模糊控制演算法的輸入，由專家知識創建的模糊推論引擎判斷行為模式，使投餌機器人行駛時能自主的調整舵角度並保持與牆壁為水平狀態。其中電子羅盤的方位角結合慣性感測單元，將載體座標透過尤拉座標轉換為導航座標，使電子羅盤傾斜時也能精確的計算方位角，而慣性感測單元的系統內部使用UKF演算法處理感測器的雜訊誤差，解決陀螺儀隨著時間而產生累積誤差的影響。 投餌機器人的定位方式使用GPS的都普勒效應(Doppler)，由速度資料透過梯形積分估算移動距離，GPS結合電子羅盤的轉向角度，將行進的距離和轉向角度透過座標轉換為(X,Y)定位座標，定位的用途在於開啟投餌機設備的量測依據。因為投餌的飼料密度攸關著蝦子的成長，過量的投餌會造成水質汙染，過少則會讓蝦子生長緩慢。所以投餌機器人需定位出於養殖池中所移動的距離，並依照移動固定距離後開啟投餌機設備，投餌機依照系統的設定，機器人沿牆行進固定150公分後開啟投餌機一次並關閉，待下次行進150公分後再開啟一次，主要是將飼料均勻地散播於養殖池中。

This research was to develop an integrated localization and navigation control system of an autonomous feeding robot for shrimp farming. Feeding is a very important process in shrimp farms. An autonomous feeding robot could feed shrimp 24 hours a day to alleviate restrictions by available human labor. The feeding robot must be able to perceive the environment and its own position. While the feeding robot should follow the bank of the shrimp pond, the navigation system should accurately measure the azimuth and bank distance. Two ultrasonic sensors are mounted on a motor-driven base to enable automatic tracking of the pond bank. Therefore the distance of the feeding robot to the bank could be accurately determined even when the robot is more than 45 degrees from the bank. A fuzzy control algorithm using azimuth and bank distance as input guides the feeding robot by adjusting the angle of the rudder to keep the robot to navigate parallel with the bank. Azimuth readings from the electrical compass combined with signals from an inertial measurement unit (IMU) are used to transform the robot local coordinates into global coordinates via Euler coordinates transformation. The electrical compass could accurately calculate azimuth even if it tilts. The IMU uses an UKF algorithm to reduce noise of the gyroscope sensor to eliminate accumulated error. The localization method uses the built-in Doppler function of a GPS to estimate the moving distance of the robot via trapezoidal integral. The result is then combined with the turning angle from the azimuth to determine the x-y coordinates of the robot. The accurate localization of the feeding robot ensures that a spreader on top of the robot could deliver feed to the shrimp pond every 150 cm or set distance to evenly spread the feed in the shrimp pond for best shrimp growth.