【作者】 赵迪；

【导师】 李世其；

【作者基本信息】 华中科技大学 ， 机械设计及理论， 2011， 博士

【摘要】 机器人遥操作能够代替人在远距离、真空、高温和高辐射等人类难于直接触及的环境下从事科学实验、装配作业和维护作业等。由于距离的影响,遥操作中的时延常常是变时延。变时延下的机器人遥操作,关键难点在于如何克服变时延对遥操作的影响。本文研究了变时延下机器人遥操作的若干关键问题。首先从虚拟环境建模和动力学建模两方面,对机器人遥操作系统的构建进行了分析。实体建模到运行仿真,为机器人遥操作系统的建立了平台。在建模过程中考虑了仿真运行的效率问题。采用模型简化的方式,既从视觉上保证了不影响临场感效果,又从物理上不影响仿真执行结果。针对机器人遥操作任务复杂的问题,提出了将复杂遥操作任务进行结构化描述的方法。以同构的六元组结构来描述遥操作任务,便于计算机识别和管理,充分利用从端机器人的自主智能。对复杂遥操作任务进行面向对象的编程和任务分解,并将遥操作任务的操作过程划分为“接近-操作-离开”三阶段。所有的任务均按这种阶段方式进行划分,直至划分到不可再分的“动素”级别。为了在任务级操作过程中加强主端的干预能力,提出了任务级遥操作的切换策略,避免从端在任务执行过程中遇到困难死锁或误操作。提出了将滚动预测用于从端机器人状态预测的方法,利用BP神经网络对机器人的从端当前状态进行预测,减少因时延带来的反馈信息滞后所造成的“运动-等待-运动”操作形式。仿真实验表明这种基于BP神经网络的滚动预测可以有效的减少预则仿真的误差。针对遥操作中力反馈信息滞后和反馈信息采样率低的问题,提出了采用真实材料特性接触力曲线辅助虚拟力计算的方法,提高虚拟力计算的效率；提出虚拟力外推的方法,提高力反馈信息采样频率,以获取平稳的力反馈效果。仿真实验证明,该方法易用、可行。在遥操作回路中引入波变量,以平衡时延对遥操作系统的影响。设计了时延模似器,在局域网对遥操作时延环境进行模拟。在对时延进行有效估计的前提下,基于无源性理论对遥操作系统稳定性进行了分析。能够证明系统在波变量的辅助下可以实现稳定遥操作。最后,开发了“武汉-哈尔滨”机器人远程遥操作系统,在5-7秒变时延条件下进行了四项实验操作。综合验证了任务级遥操作的方法、虚拟力外推算法和滚动预测仿真。实验表明这种遥操作系统可以充分的利用机器人的自主能力应对较为复芬的遥操作任务,又减少了操作者的工作强度和保证了主端操作的临场感。更多还原

【Abstract】 Teleoperation robot is capable to work in some terrible environments or in place which is difficult to reach for human, such as long distance, vacuum environment, high temperature environment and radiation environment etc. Scientific experiments, assembly operation and maintenance work can be done by tele-robot. Time delay is varied by time because of the long transmission distance. One of the key difficulty to long distance teleoperation is to reduce the effect of time-vary delay. This dissertation studied some key technologies of time-vary delay teleoperation to robot.Firstly, how to establish a teleoperation system to robot is discussed by two branches, Virtual Reality (VR) environment modeling and dynamics modeling. The VR environment is built to simulation movements of the objects. The efficiency of he simulation must be considered. So the model has to be simplified before VR environment is built. Both the immersion and physical characteristics are required.Aimed to the complicated tasks to robot teleoperation, a task description method is put forward. A six-elements-structure established to descript teleoperation task, and every level of task use the same kind of structure. The structure can be easily read by computer programme, and autonomous intelligent of slave robot can be used to help finish the complicated task. Object oriented method is used to programe teleoperation task and to break down the task. Teleoperation task can be break down to three stages which are "reaching", "operation" and "leaving". Every level of task and every progress is break down to this three stages, until the therbligs level. The switching policy is promoted to increase the intervention of maseter side during the operation. This policy can help slave side to avoid some mistakes.A rolling prediction method is promoted to help predict the state of slave robot. BP Artificial network is used to predict the online state of slave. This prediction can reduce "move-wait-move" operation which is brought by time delay. Simulation experiment shows that this rolling prediction method can decrease the error of predictive simulation.Aimed to delay of force-feedback information and the low sampling rates of force sensor, a virtual force calculation method is put forward. This method is based on contact force database of material. An extrapolation algorithm is used to increase the sampling rates of force feedback. Simulation indicate that this method can gain smooth force feedback.Wave-variable is imported to reduce the affect to the stability of teleoperation system. Time delay simulation software is programmed to simulate time-vary delay environment in intranet. The stability of the teleoperation system is analysed by passive theory.Finally, a "Wuhan-Herbin" long distance robot teleoperation system is developed. Four experiments are finished under 5-7 seconds time-vary delay. Task level method, virtual force extrapolation algorithm and rolling predictive simulation are tested. Experiment indicate that this teleoperation system can use autonomouse intelligent of slave robot to finish complicated teleoperation task, provide immersion operation and reduce the working strength of human operator in master side.更多还原