- 學位論文
智慧演算法於四足機器人平衡與運動控制
Intelligent Algorithm for Balance and Motion Control of Quadruped Robots
摘要
本論文成功地利用STM32F446 Nucleo實驗板與Raspberry Pi 4樹莓派開發板，實現了一並聯式機構設計方式之四足機器人，其中STM32F446 Nucleo實驗板主要負責機器人I/O介面控制與運動控制，Raspberry Pi 4樹莓派開發板主要負責機器人影像處理工作。同時，參考四足生物之步態，本論文規劃設計出一套適合所提出四足機器人之步態規劃，並透過逆向運動學可以求得各馬達所需之角度命令。接著，在控制器設計部份，分別提出了平衡控制器、避障控制器及循線控制器等三種不同控制目的之智慧型演算法，平衡控制器可以透過姿態感測器即時量測地面傾角後，改變機器人四隻腳之長度，使得機身可以保持水平；避障控制器可以透過超音波感測器即時量測與前方障礙物距離後，讓四足機器人能夠自主的前後移動與障礙物保持固定距離；循線控制器則先透過影像處理技巧即時感知地面黑線的變化後，控制機器人左右兩側之行進速度不同來讓四足機器人能夠跟著黑線行進。最後，經由實際的實驗結果展示出，所提出四足機器人不只可以擁有良好的移動能力，還可以結合智慧型演算法得到不錯的控制效果。
並列摘要
This thesis successfully uses STM32F446 Nucleo development board and Raspberry Pi 4 computer to implement a quadruped robot with parallel mechanism design, where STM32F446 Nucleo development board is mainly responsible for I/O interface control and motion control and Raspberry Pi 4 computer is mainly responsible for robot image processing. At the same time, this thesis designs a set of gait planning with reference to the gait planning of quadrupeds, and the motor control commands can be obtained through inverse kinematics. Then, three intelligent controllers for different purposes are proposed, which are named as the balance controller, the obstacle avoidance controller, and the line-following controller respectively. The balance controller can measure the inclination of the ground through the attitude sensor, and then change the length of the four legs of the robot so that the body can be level. The obstacle avoidance controller can measure the distance to the obstacle through the ultrasonic sensor, so that the quadruped robot can move forward and backward autonomously to keep a fixed distance from the obstacle. The line-following controller can sense the changes of the black line on the ground through image processing techniques. Using to control the different travel speeds of the feet on the left and right sides of the robot, the quadruped robot can automatically follow the black line. Finally, the experimental results show that the proposed quadruped robot can not only have good mobility but also can achieve favorable control responses with using intelligent algorithm.