PID控制器與自調適輸出預測補償在四軸飛行器影像伺服的應用研究

Abstract

以視覺伺服追蹤目標並進行載具控制時，視覺影像傳輸過程因取樣速度及資料量龐大會造成延遲，降低系統響應頻寬，而無線影像傳輸速率低於有線傳輸；響應較慢的載具如船舶、飛行器相較於汽車，需要更多的時間來達成指定的控制，造成追蹤影像目標的困難。本研究以局部二值模式串聯分類器進行快速的影像識別，加入預測目標物未來位置的演算法，搭配PID控制器來進行影像伺服(Visual Servo)控制，藉此降低位置影像傳輸延遲帶來的負面影響，同時加快載具的反應速度。本研究以四軸飛行器為載具，使用搭載的攝影機為感測器，將影像以無線傳輸方式回傳到電腦進行影像處理，再由電腦計算預測位置，並下達飛行指令，藉此優化系統響應頻寬及驗證以輸出預測補償強化的PID控制。此演算法有效地改善因位置影像傳輸延遲迴授造成即時控制失誤之性能，增加四軸飛行器控制的穩定性及響應速率。

Vehicles controlled by visual tracking will encounter image transmission delay, especially in wireless case. Some mobile vehicles like ships and aircrafts will need more time to respond than cars, cause difficulties in tracking. This study utilize LBP (Local Binary Pattern) cascade classifier to detect target and a PID controller with prediction algorithm to minimize the negative effect caused by delay. This study use quadcopter as vehicle with its camera to capture image, and utilizing wireless protocol to transmit video to ground station computer. Computer carries out collecting data, performing the calculations, verify and optimize the output prediction with enhanced PID control, and then send control parameters back to quadcopter. This algorithm has reduced the negative impact brought by delay, thus enhance quadcopter's stability and response.