| 摘要: | 本研究主要精神是仿效生物對規律性動作的控制機制,其運作是由一群低階神經細胞經互相傳遞及影響,產生規律的周期訊號,再藉由外界刺激與大腦刺激修正該訊號,共同合成為最後運動的型式,此即為中樞型態產生器 (Central Pattern Generator)。本系統將以 Matsuoka′s 神經振盪器作為控制六足機器人動作的 CPG(Central Pattern Generator) 基本組成單元,並搭配三軸加速度計與三軸陀螺儀即時獲取機器人姿態,利用此姿態資訊當作CPG 的回授訊號,使其在中度不規則地形能夠回復水平姿態。仿 CPG 模型控制器為一種分散式控制方法,每個足部的控制器由一群神經振盪器與感受單元所組成,每個足部的控制器彼此互相耦合,藉由不同的耦合方式產生出不同的運動步態。整體控制架構經由個別足部方向的傾角作為回授訊號輸入到神經振盪器改變其振幅大小,再與固定振幅的參考振盪器做比較產生出能平衡身軀的足部高度參考訊號,隨後將此控制訊號經由軌跡產生器轉換為機器人足部動作之軌跡,此軌跡再經由逆運動學得到實際的伺服馬達轉動角度以控制馬達轉動角度,進而達到行進時同時恢復平衡的效果。
;In this thesis, we built a controller that can produce rhythmic locomotion inspired by biology concepts of CPG(Central Pattern Generator). CPG, a group of lower-level neural oscillators coupled each other in the creature’s central system, can produce coordinated oscillatory signals for rhythmic locomotion. The oscillatory signals can be modulated by sensory stimulator or higher-level input. We used the Matsuoka′s neural oscillator to construct CPG-based controller of the Hexapod walking robot for generating feet control signals. The body attitude, used for feedback pathway, can be captured by fusing IMU’s data from 3-axis accelerometer and 3-axis gyroscope. Hexapod robot can be up to horizontal when walking on terrains of medium degrees of irregularity by using the CPG as a controller and body attitude as feedback. The CPG is a distributed control method. Each foot has their own CPG controller construct by a group of neurons and sensory feedback. The different output pattern can be produced by different topology of the CPG. The tilt of each foot’s direction is used as feedback for the external neural oscillator to change oscillation amplitude. Then produce the control signal by comparing the external neural oscillator and reference neural oscillator. Inverse kinematics is used for converting the control signal to produced motor angle command. The cooperation of each part of this system achieved to make hexapod robot walking with tilt recoverability. |
