- 學位論文
具備全方位視覺之全方位機器人的自我定位方法與發展平台
Self-localization Method and Robotics Platform of Omni-directional Robots with Omni-directional Vision
摘要
本論文針對具備全方位視覺之全方位行動機器人，發展自我定位方法與發展平台。機器人自我定位方面，使用粒子過濾器以機率分佈方式搭配機器人的運動模組與感測模組，遞廻地求算機器人在環境中的位置狀態。運動模組是由馬達編碼器結合機器人運動學所構成，感測模組則是以全方位視覺偵測影像特徵點與計算影像深度所構成。機器人平台發展方面，建立機器人機電整合的軟體模組，解決具備不同硬體介面之機器人共同發展的問題，包括人形機器人、兩輪行動機器人、與全方位機器人等。針對視覺、無線通訊、與運動控制器等介面溝通問題，提出解決方案。機器人平台將架構在Windows PC-based 控制器上。
並列摘要
In this thesis, a self-localization algorithm and a software platform are developed for an omni-directional robot with omni-directional vision. The research is divided into two parts: robot self-localization and robotics software platform. In the development of robot self-localization algorithm, a recursive particle filter with probabilistic distribution is utilized to find the robot pose in the environments. The particle filter is composed of a motion model and a sensor model. The motion model is constructed based on robot kinematics with motor encoder feedback, while the sensor model is established by using an omni-directional vision system. In the development of robotics software platform, a software module is programmed for the robot mechatronics system to solve the interface problem for robots with heterogeneous hardware, including humanoid robots, two-wheeled robots, and omni-directional robots. The software platform is developed on Window PC-based controllers to resolve the interface and communication problems between these controllers.