一种气动静压软体机器人的驱动力产生机理及控制策略  

The Driving Force Mechanism and Control Strategy of a Pneumatic Hydrostatic Soft Robot

作　　者：郑俊君 宋小波[2,3] 姜祖辉[1,2] 谭治英[2,3] 

机构地区：[1]中国科学技术大学,安徽合肥230031 [2]常州先进制造技术研究所,江苏常州213164 [3]中国科学院合肥物质科学研究院,安徽合肥213016

出　　处：《机器人》2014年第5期513-518,共6页Robot

基　　金：国家自然科学基金资助项目(61105111)

摘　　要：基于复杂刚性结构的传统机器人在狭小多变的空间中活动时,易对机器人本体产生较大磨损与消耗.针对此类问题,本文提出一种新型气动静压软体机器人,降低了机器人通过狭小空间时的阻力,减少运动损耗.在研究海参静水骨骼结构基础上,设计了软体机器人气囊相变结构,分析了气囊充放气的流体静压力学特性,建立驱动力产生的数学模型,从而设计出单单元与双单元控制策略,并利用时间戳BP(反向传播)神经网络对系统的延时时间进行预测,增强内模反馈系统的稳定性.在越障实验与稳定性分析中,加入延时预测后,机器人能在单单元与双单元控制模式下稳定行进,并顺利越过小沟壑,顺利完成六单元翻转的概率从89%提高到96%.Traditional robots with complicated rigid structure will be worn and teared when running in a cramped and diverse space. For this problem, a new model of pneumatic hydrostatic force soft robot is presented, and the resistance and motion loss are reduced while robots pass the cramped space. Based on sea cucumbers＇ hydrostatic bone structure, a soft robot with airbag phase transition structure is designed, the hydrostatic mechanical property of the airbag＇s charging and discharging process is analyzed, and a mathematical model of driving force is set up. Thus, single-unit and double-unit control policies are designed, meanwhile, a time-stamped BP （backpropagation） neural network is used to predict delay time for increasing the stability of internal model feedback control. According to obstacle surmounting tests and stability analysis tests, the robot with delay prediction can move stably and cross small gaps smoothly by single-unit and double-unit control strategy, and the probability of turning over six units increases from 89% to 96%.

关 键 词：软体机器人 流体静压特性 控制策略 延时预测 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]