應用於四輪移動機器人車的動態避障系統

Abstract

本論文藉由整合四輪驅動車與機械手臂，完成一台移動機器人車。我們設計一結合了光流法與 SVM 分類器的移動行人影像運動偵測系統，以實現移動機器人的動態避障功能。此外，亦設計了一僅以單一影像輸入的影像伺服控制系統，用以精確的控制機械手臂完成夾取作業。最後，整合上述兩項功能，使移動機器人可以在複雜的工作環境中避障移動以完成夾取作業。移動機器人車的移動速度控制功能是由模糊控制器串聯比例、積分及微分控制器(Proportional-Integral-Derivative, PID)在微控制器中實現。應用於機械手臂的影像伺服控制系統利用單目測距，以單一組攝影機提供的影像輸入計算出目標物件的世界座標。將此資訊回傳至微控制器後，由微控制器計算並控制機械手臂移動至夾取物體的姿態。最後，移動機器人車透過整合實驗，驗證此機器人可以完成夾取指定物件，並在移動過程中對於行人進行避障的任務。This thesis designs a mobile robot vehicle that combines a four-wheel vehicle and a robotic arm. We design a image motion detection system to capture moving pedestrians. This system is based on optical flow method and SVM classifier, and is applied as the input of the dynamic obstacle avoidance function. In addition, we also design an image servo control system that only uses a single image input to precisely control the robotic arm to complete the gripping operation. Finally, by integrating the above two functions, the mobile robot can avoid obstacles and complete the gripping operation while moving in a complex working environment.The movement speed control function of the mobile robot vehicle is completed by a fuzzy controller in series with a PID controller implemented in the microcontroller. The image servo control system uses a single input image to extract the coordinate of target objects with monocular ranging. After sending the coordinate to the microcontroller, the microcontroller calculates and controls the robot arm to move to the posture of the gripping object.Finally, this mobile robot vehicle verifies its object gripping and obstacle avoidance function through several experiments.