輪爪變形機器人之爪式爬階性能與輪式操控策略

Abstract

本論文旨在提升輪爪變形機器人於爪式運動模式時的爬階性能，以及輪式運動模式時的操控表現。此研究延續團隊先前所發展之輪爪變形架構，針對軟、硬體提出創新與改良。改良後的折疊轉換機構使得變形過程大幅簡化，只需要抬升與降下前機體後，保持爪形或是形成合成輪即可，轉換過程較不受地形因素影響。在分析前代二階跨距機器人之幾何規格以及其爪式爬階性能後，提出加長機體長度的三階跨距機器人，以改變質心與接地支持多邊形之相對位置，大幅提升爬階穩定度。此外，大幅縮減前後馬達扭矩之差異，讓馬達負載更為均衡。在輪式運行中，特殊的合成輪以接地輪爪和懸空輪爪形成，兩者分別以轉速控制符合差速運行之移動速度與迴轉半徑，以及角度追蹤控制保持合成輪成立，最後加入航向角誤差回授機制，能更精準地控制機器人在實際環境之方位。由爬階性能模擬可得，三階跨距機器人較二階跨距機器人之爬階性能較佳。在輪式轉向模擬中，根據設定的移動速度與迴轉半徑，可得到接地輪爪轉速值以及週期比。由爪式轉換至輪式實驗中，可證實折疊轉換過程之可行性，並較前代系統之操作簡單。在爬階實驗中，機器人依照爬階策略可更穩定地攀爬階梯。於輪式運行實驗中，加入航向角誤差回授機制後，機器人可更準確地追蹤目標航向，並大幅減少與預設路徑之誤差。

This thesis develops the Claw-Wheel transformable robot that enhances the stair climbing performance in claw mode and maneuverability in wheel mode. The concept follows the Claw-Wheel transformable structure as the research group before, and focus on the innovation and improvement of the hardware and software. The improvement of the transformation mechanism simplifies the transformation process, which just lifts the front body up and lay the front body down to the ground, and forms the claws or composite wheels; the process would not be much affected by the rough terrain. After analyzes the geometric specification and stair climbing performance of the robot striding two steps in last version, this research develops the robot with extended body length for striding three steps. Thus, the relative position of the mass of center to the support polygon is changed, and the stability of stair climbing is more enhanced. Besides, the difference of the required torques between the front and rear motors could be narrowed down and more balanced. In the movement of the wheel mode, the driving wheels are formed with the claws in stance and claws in flight; the claws in stance are controlled by the speed control for following the speed of movement and radius of gyration in differential drive, and the claws in flight are controlled by the angle-tracking control for forming the driving wheels. Finally, the switching control strategy with yaw feedback make the robot track the desired orientation more correctly. In the simulation of stair climbing performance, the robot striding three steps has better climbing performance than striding two steps. In the simulation of differential drive, the movement of speed and radius of gyration are set, and the angular velocities and period ratio of the claws in stance can be computed. The experiment of the claw mode transforming to wheel mode verifies the feasibility of transformation process, and the process is more simplified than before. In the experiment of the stair climbing, the robot climbs the stairs more steadily under the stair climbing strategy. In the experiment of the differential drive, after adding the yaw feedback, the robot tracks the desired yaw angle more accurately, and reduces the errors of the expected path.