新型輪腳複合機器人研發與其爬樓梯策略

Abstract

本研究提出一款新型輪腳複合機器人Kilin，針對現存輪式機器人在面對非連續性樓梯時的穩定問題，自構型與機構設計的角度為出發點，以實際樓梯尺寸來設計機器人尺寸，於機構設計中以蝸輪蝸桿齒輪箱與花鍵槽軸的機械特性，展現具備可自鎖之快速運動模式切換足模式與輪模式的模組化解決方案。本設計以實用性與控制簡化為導向，整合簡化版的四足輪腳架構與四輪獨立驅動與轉向能力，使機器人在室內外平地保持輪式機器人的優勢之外，在面對樓梯時仍能展現越階能力，穩定運行。為驗證本設計之可行性，本研究亦提出基於Kilin構型的爬樓梯策略，將複雜的爬樓梯任務階段式拆解為數個不同目的之子狀態流程，搭配馬達狀態訊息回授進行自動狀態切換，不需要提前知道樓梯階數，即可完成爬樓梯任務。透過實體實驗與分析，包括不同階高與階深組合下之多次重複測試，檢驗在簡化構型設計、機構設計下之爬樓梯策略的穩健性，並以俯仰角之減少程度與穩定度來驗證設計合理性。最終結果顯示，所提出之設計與策略具備越過多種單階台階與多階樓梯之實際應用潛力，展現良好的樓梯適應性。

This study proposes a novel wheel-leg hybrid robot named Kilin, addressing the stability challenges faced by conventional wheeled robots when encountering discontinuous stair environments. Starting from a design perspective focused on reconfigurability and mechanical structure, the robot's dimensions are determined based on actual stair measurements. The mechanical design incorporates a self-locking worm gear reducer and a splined shaft mechanism to enable rapid and modular transitions between legged and wheeled locomotion modes. Prioritizing practicality and control simplification, the robot integrates a streamlined quadruped wheel-leg structure with independently driven and steered wheels. This design retains the advantages of wheeled mobility on flat terrains while demonstrating stair-climbing capability and operational stability in indoor and outdoor environments.

To validate the feasibility of the design, this study also proposes a stair-climbing strategy tailored to the Kilin architecture. The complex stair-climbing task is decomposed into several sub-states, each with specific goals, forming a staged finite-state process. Automatic transitions between states are achieved through motor feedback without requiring prior knowledge of the number of stair steps. A series of physical experiments and analyses—across various step height and depth configurations—are conducted to evaluate the robustness of the climbing strategy under the simplified mechanical design. The degree of pitch angle reduction and system stability are used as key metrics to verify the effectiveness of the proposed approach. Experimental results demonstrate that the proposed robot and control strategy possess practical potential for negotiating both single and multiple stair steps, showcasing strong adaptability to stair-climbing tasks.