本論文提出一個基於機器人作業系統(Robot Operating System, ROS)之人形機器人的實現方法，ROS為分散式的架構，並以點對點網路將所有的處理序連接在一起交換資訊。本論文在Linux環境下以ROS建構人形機器人的軟體開發系統，並與嵌入式系統SoC FPGA整合，實現軟硬體協同設計。本論文實現人形機器人的行走速度規劃，以人形機器人的腰部作為質量中心點(Center of Mass, CoM)，透過規劃CoM的速度規劃產生雙足步態軌跡，再透過雙足運動模型計算關節旋轉角度，使人形機器人具有行走能力。在行走過程中，因為馬達背隙、地面不平或是機構重量因素，造成人形機器人在運動過程中不平穩甚至跌倒，因此本論文加入卡爾曼濾波器(Kalman Filter)，融合陀螺儀與加速度計估測出身體傾斜角度姿態，再透過姿態控制器計算出角度補償，使行走過程中不會跌倒。最後，結合控制實際的人形機器人與Gazebo模擬器，實現虛實整合系統。 This thesis proposes an implementation of humanoid robot based on Robot Operating System (ROS). The system architecture of ROS is a distributed system. The ROS uses peer-to-peer network to link all processes to exchange data. In the Linux environment, the humanoid robot system is built to develop the software system through ROS which combines with embedded system SoC FPGA and achieve hardware and software co-design. This thesis proposes speed planning of humanoid robot. The waist of humanoid robot to be the Center of Mass (CoM). According to speed planning of CoM to produce the walking trajectory of biped. And then motor angle is obtained by kinematics module to make humanoid robot to walk. Because of gears backlash, uneven floor, and mechanism weight, it made humanoid robot unstable or even fall while walking. Therefore, the system uses Kalman filter that combine gyro sensor with acceleration sensor to estimate the slope angle of upper body. Then posture controller is used to determine an angle compensation to make humanoid robot walk stable. Finally, this thesis combines reality humanoid robot and simulator Gazebo to achieve Cyber-Physical System (CPS).
