本論文針對視覺自主之人形機器人設計與實現足球守門員策略。在Linux環境下,以機器人作業系統(Robot Operating System, ROS)建構人形機器人守門員之軟體策略架構。守門員策略依據人形機器人位置、球位置、人形機器人與球的距離、球的移動方向為條件,策略執行流程分為兩個部分:(1)守門員定點找球和(2)守門員防守行為。在守門員定點找球部分,可以分為頭部尋球以及頭部追球共兩個動作,規劃轉動頭部攝影機的流程,目標為找到球並持續追蹤。在守門員防守行為部分,可以分為對正目標移動、平移防守、蹲下防守、踢球和走回球門中央共五個動作,根據人形機器與球門距離訂定防守範圍,不同防守範圍對應不同動作,目標為阻擋球滾進球門之路線與踢球遠離球門。在實驗結果上,以雷射測距儀輔助紀錄小型人形機器人行走軌跡,驗證本論文提出之策略流程之人形機器人能完成基本之足球守門員策略。
In this thesis, a soccer goalkeeper strategy for vision-based autonomous humanoid robot is designed and implemented. Robot operating system is used to construct a software architecture of goalkeeper strategy of the humanoid robot in Linux platform. The goalkeeper strategy is based on the humanoid robot position, soccer position, the distance between the humanoid robot and soccer, and the soccer moving direction. The strategy execution process is divided into two parts: (1) goalkeeper place-search and (2) goalkeeper defensive behavior. In the goalkeeper place-search, it is divided into two phases: head searching and head tracking. A process of turning the head camera is planned to search and keep tracking soccer. In the part of the goalkeeper defensive behavior, it is divided into five phases: aligning and moving towards the target, moving sideways defense, squatting defense, kicking, and walking back to the center of the goal. The defensive range is set according to the distance between the humanoid robot and the goal. Different defensive ranges correspond to different phases. The purpose is to obstruct the route of the soccer and kick the soccer away from the goal. In the experimental results, laser rangefinder is used to assist in recording the walking trajectory of a small humanoid robot and verify the strategy process proposed in this thesis.