小型人形機器人之即時行走的SOPC設計與實現

SOPC Design and Implementation of Real-Time Walking for Small-Sized Humanoid Robots

劉智誠(Chih-Cheng Liu)

指導教授 ： 翁慶昌

淡江大學/工學院/電機工程學系博士班/博士(2014年)

https://doi.org/10.6846/TKU.2014.00221

摘要

本論文提出一種雙足行走步態的設計方式，並用座標旋轉數位計算器（CORDIC）演算法將行走步態和逆運動學計算利用SOPC方式實現。在行走步態方面，本論文在基於振盪為基礎的簡單模型中，提出一些可以利用正弦函數來描述雙足行走步態的公式。為了更自然更容易定義所描述的振盪器參數，本論文選擇以機器人的腰部擺動長度、步伐長度、抬腳高度以及行走時間作為參數設定，可針對機器人特性與需求各別設定，不需要精確的數學模型，即可輕易地設計所需的步態軌跡，再利用逆運動學求得步態軌跡與各馬達間的關係。由於CORDIC只需要加減法、位移及少數的記憶體即可實現根號、三角與反三角等函數的計算，所以本論文以CORDIC演算法來實現機器人之步態軌跡和逆運動學的函數計算。由實驗結果得知，借由FPGA硬體電路高速平行運算的特性，可以大幅降低IPC工業電腦與Nios II軟體執行的運算負擔，也讓IPC與Nios II較有餘力執行影像、策略以及智慧型演算法等功能。以FPGA硬體電路實現機器人的行走步態系統和逆運動學，其計算只產生小範圍的誤差。因此，本論文以CORDIC演算法將機器人步態軌跡和逆運動學利用SOPC方式實現可以達到即時控制，最後再加上利用FPGA方式實現機器人行走之地面傾斜估測方式來判斷是否行走在傾斜地面。

關鍵字

雙足機器人 ； 座標旋轉數位計算器 ； 可程式規劃系統晶片 ； 行走步態

並列摘要

In this thesis, a design method of biped gait for humanoid robots is proposed.
A CORDIC-based SOPC system is used to design biped gait and inverse
kinematics. In the biped gait, some equations represented by sinusoidal
functions are proposed to describe a biped walking. In order to be more
natural and of more practical importance, oscillation parameters are described
by the swing length, step length, and lifting height of a biped robot. These
parameters can be set from characteristics and requirements for biped robot.
Accurate mathematics model is not necessary. The biped gait can be easily
designed. Then we use inverse kinematics to obtain the relationship between
gait trajectory and each motor. Because CORDIC algorithm only needs
addition and subtraction, displacement, and mininal memory for calculating
the square-root, trigonometric, and inverse trigonometric functions, thus the
CORDIC algorithm method is used to calculate walking gait and inverse
kinematics for a biped robot in this paper. From the experimental results,
FPGAs are capable of high- performance parallel computing, the
computational burden of IPC and Nios II soft-core processor can be greatly
reduced. Let master controller can execute strategic, image processing,
intelligent algorithms , etc. FPGA hardware method is used to compute the
walking gait and inverse kinematics. These results computed by the FPGA
hardware method are almost the same as that computed by the software
method. Therefore, the proposed CORDIC-based FPGA hardware design
method can really be applied to improve the real-time control performance of
biped robot. And a estimation method of a walking robot on a horizontal plane
is designed on the FPGA hardware method to determine whether the robot can
working in slope surface

並列關鍵字

Biped Robot ； CORDIC ； SOPC ； Walking Gait

參考文獻

[2] Y.T. Su, K.Y. Chong, and T.H.S. Li, “Design and implementation of fuzzy policy gradient gait learning method for walking pattern generation of humanoid robots,” International Journal of Fuzzy Systems, vol. 13, no. 4, 2011, pp. 369–382.

[29] C.C. Wong, C.T. Cheng, K.H. Huang, and Y.T. Yang, “Fuzzy control of humanoid robot for obstacle avoidance,” International Journal of Fuzzy Systems, vol. 10, no. 1, 2008, pp. 261–270.

[1] J. Perry, Gait Analysis: Normal and Pathological Function, Downey, CA: Rancho Los Amigos Medical Center, 1992.

[3] W. Yang, H. Kim, and B.J. You, “Biologically inspired self-stabilizing control for bipedal robots.” International Journal of Advanced Robotic Systems, vol. 10, no. 144, 2013.

[4] M. Vukobratovi’c and J. Stepanenko, “On the stability of anthropomorphic systems,” Mathematical Biosciences, vol.15, Oct., 1972, pp.1–37.