以具能量輸入和損耗之單足滾動模型建構多足機器人在黏彈地表上之跑步步態

Abstract

隨機器人產業逐漸走進人類社會，逐步開始負擔更加複雜工作，其對於自由在各種地形自由活動之需求也隨之提升。為達成此目標，結合生命科學領域之仿生機器人科技正蓬勃發展，藉由觀察生物之行為並以之發展簡易模型做為控制模板，期許能賦予機器人更靈活的行動能力。本論文從實驗室建立之具有滾動特性之R-SLIP模型、以及加入阻尼及扭力造成之能量流動之TDR-SLIP模型出發，藉由從能量觀點進行之模擬、以及大量實驗探討利用模型控制機器人產生高動態步態之可行性。在成功達成高速動態步態的實現後，另一難題是加入與地面的交互互動。生物於行走時，仰賴複雜的感官與肌肉協調，而能在各式崎嶇軟硬程度不一之地面行動，但在機器人身上要達到同樣程度需要複雜運算與高強度機構而有相當困難性。本文發展出藉由電流控制馬達扭力之複合控制架構，希望能使機器人有感測力之能力，並在TDR-SLIP模型中加入地面模型，以模擬和實驗探索會變形之地面對運動所帶來的影響。本研究目標為初步探討高速運動時地面與足部之交互關聯，除了希望能以此模型探討生物行走時足部與地面之關聯，並提供將來進一步發展更為完善的地面模型之基礎，期許最終能賦予機器人於各種地面皆能同時高速且穩定的以動態行走之能力，使機器人的活動範圍拓展至更寬廣的範圍。

In recent decades, robots have gradually taken on more tasks that require mobility. Unlike animals, which are capable of coordinating complex sensing organs and muscles to achieve dynamic motion on all kinds of terrains, robots have difficulties traversing rough terrains at high speeds due to limitations of computational resources, actuator strength, etc. The aim of this research is to build a new model incorporating the interaction between the leg and the ground based on the previously developed dynamic model, which can achieve dynamic running on rigid ground. Extensive experiments were conducted to examine how the rolling contact and energy-flowing characteristics of the TDR-SLIP model interact with one another in the process of dynamic motion. Following this work, a position-torque hybrid control method was developed to enable the robot’s force-sensing ability. Then, in order to determine the effect of the leg-terrain interaction on the dynamic motion, a new model was developed, in which the ground is described by a combination of linear springs and dampers and which can model a wide range of terrains. Simulation results show chaotic behavior, and such behavior was verified by an experiment. This work provides the basis for future research on how the ground impacts the locomotion of the animal and can serve as a foundation for the building of better terrain models, which may make robots capable of dynamically traversing on all kinds of terrain.