Omni Planar Direction Push and Recovery System in Mimicking Human Actions for Humanoid Robotics

隨著機器人領域的蓬勃發展，各式的機器人研究也愈來越廣泛。其中雙足機器人和人形機器人在機器人領域裡又占據了很大的一部分。人形機器人在人類日常的活動環境相較於其他種類的機器人有較高的適應性和較高的靈活度。但是相較於輪型機器人與多腳型機器人，維持類人形機器人的穩定度是一大挑戰。如何在靜態和動態時維持平衡和穩定是類人形機器人領域裡最優先需要處理的議題。 在人類活動的各式環境裡，類人型機器人的活動不可避免的會與外在的物體或環境發生碰撞。如果類人形機器人將來要走入人類的生活提供服務，受到外力干擾並復原是首先要解決的問題。有了外力干擾及復原系統的機器人才能在日常生活的環境裡安全的行動並完成各式任務。 本論文研究旨在將外力干擾回復系統整合進入類人型機器人步態軌跡產生器中，使機器人在動態行走時能維持平衡並保持穩定度。為了實現外力干擾復原系統，我們觀察並模仿人類在面對外力干擾時的反應並以此為概念來建構此系統。 在本研究中，我們成功地將模仿人類反應的外力干擾復原系統實現於國立臺灣大學智慧機器人及自動化國際研究中心實驗室研究發展的人形機器人上。並將此系統整合於機器人的線上步態軌跡產生器中，使得類人形機器人在行走過程中即使受到外力的影響仍然能夠維持行走時的穩定度。本研究包括類人形機器人理論的推導，並實際行走並受外力干擾後回復平衡驗證此外力干擾回復系統的實務性及可行性。期望本研究能夠在類人形機器人行走的外力干擾復原研究上提供學術及實務應用之貢獻。With the rapid development of the field of robotics, the research field of the biped and humanoid robots has increase extensively. The bipedal and humanoid robots require higher elasticity to move in the complex environment than other kinds of robot. Howev-er, unlike wheeled robots and multi-legged robots, maintaining the bipedal and human-oid robots walking stability is a big challenge. The static balancing and dynamic balanc-ing problems are the most important problem in the walking stability issues. As humanoid robots begin to walk away from laboratory, start entering to the gen-eral environments. Humanoid robot is inevitable to bump into other things or human be-ings. When these unpredictable collisions occur, robot having strategy to avoid falling down and maintain balance is necessary. Consequently, the push-recovery issue has a high priority in the research fields of the humanoid robots. The research purpose of this thesis is to fulfill dynamic balancing walking by inte-grating the push-recovery system into the walking pattern generator for the humanoid robots. This system is based on the idea of the human’s behaviors when facing the ex-ternal push. In this research, we integrate the mimicking human actions push-recovery system into the walking pattern generator of humanoid robots to deal with the unpredictable external perturbations. This research consists of the theoretical derivations of humanoid robot system, push-recovery system architecture and experimental results analysis. The push-recovery system is realized on the humanoid robots in our NTU-iCeiRA laboratory. Hope that this research would contribute to the academic and practical applications in the field of humanoid robot push-recovery communities