全向移動平台結合機械手臂動態物件追蹤

Abstract

全向移動平台（Omnidirectional Mobile Platform）是一種具有全方向移動能力的移動平台，比起傳統的四輪平台更加靈活且複雜。本文自行設計此移動平台並結合機械手臂與影像辨識系統，並整合軟、硬體功能，最後使其能夠模擬一些簡單的人體動作。在機械手臂方面，描述了手臂的運動模型，取得末端的位置座標；在影像辨識上，利用雙目測距取得球體的世界座標；再將機械手臂與雙目估計的座標整合，最後透過拋物線運動方程式以及類神經網路預測其落點。最後通過實驗結果證明所提出的方法可以整合不同的座標系，且可以追蹤球體的座標，及時回傳並移動到預測落點的位置，再控制機械手臂到實際球體落下位置完成接球動作。The omnidirectional mobile platform is a type of mobile platform with omnidirectional movement capability, making it more agile and complex compared to traditional four-wheeled platforms. This thesis, designed this mobile platform and integrated it with a robotic arm and an image recognition system to simulate some simple human movements.In the field of robotic arms, the motion model of the arm is described to obtain the coordinates of the end effector. In image recognition, binocular distance measurement is used to obtain the world coordinates of the ball. The coordinates estimated by the binocular vision system are then integrated with the robotic arm's coordinates. Finally, through the projectile motion equation and neural network, the landing point of the ball is predicted. The experimental results prove that the proposed method can integrate different coordinate systems, and track the coordinates of the sphere, return it in time and move it to the predicted landing point, and then control the robotic arm to the landing position of the sphere to complete the catching action.