針對小型人形機器人，本論文設計與實現一個外力干擾回復平衡控制，主要有三個部分：(1)外力偵測、(2)跨步平衡踏點估算、以及(3)跨步平衡步態設計。在外力偵測部分，使用加速度計與陀螺儀來偵測目前機器人是否受到外力干擾。在跨步平衡踏點估算部分，借由機器人受到的外力干擾，以線性倒單擺飛輪模型來估算能夠保持平衡不跌倒的跨步策略與跨步距離；在跨步平衡步態設計部分，利用中樞型態產生器(Central Pattern Generator, CPG)以正弦函數震盪器的方式來設計與規劃跨步平衡的步態軌跡。由實驗結果可得知，當人形機器人在受到外力干擾時，所提出之外力干擾回復平衡控制可以讓機器人能夠跨出步伐並保持平衡，因此所提出的方法確實可以有效的避免機器人跌倒。 In this thesis, a push and recovery balance control is designed and implemented for small-sized humanoid robot. There are three main parts: (1) push detection, (2) calculate capture point for stepping balance, and (3) gait design for stepping balance. In the push detection, an accelerometer and a gyroscope are used to detect weather the robot is been pushed by external forces. In the capture point calculate for stepping balance, a linear inverted pendulum with flywheel model is used to calculate the capture point when the robot is been pushed by external forces. In the gait design for stepping balance, a Central Pattern Generator (CPG) is used to design and plan a gait for stepping balance with simplified sine oscillator. As the result of the experiment, the robot is able to step to the capture point and maintain balance when it is been pushed by external forces. Therefore, the proposed method can efficiently prevent the robot falling.
