基于数据驱动的微操作机器人动态不确定性建模与补偿技术研究

The rapid progress of micro-nano technology has pushed many research fields, e.g., mechatronic, manufacturing, material science, biology and medical science into the micrometer/nanometer scale level. The micro-manipulation robot is the key equipment widely used in various fields of the micro-nano technology. To image and manipulate the micro-nano object, the accurate modeling of the micro-manipulation robot system is very important for the accurate and fast positioning and control. However, various uncertain dynamics existing in the system make the first-principle based modeling very difficult. Until now, there are no mature modeling approaches for this class of systems. This project is focused on the modeling of various uncertain dynamics in the micro-manipulation robot system. By analyzing the characteristics of the uncertain dynamics and also considering a priori system knowledge, the data-driven modeling and compensation approaches for uncertain dynamics are proposed for the micro-manipulation robot system. To compensate the nonlinear, spatio-temporal coupled uncertainties and inaccurate system parameters of the micro-cantilever, an intelligent spatio-temporal modeling approach is proposed. To compensate the multi-feature coupled uncertainties on the creep, vibration and hysteresis of the piezo-actuator, a multi-feature integration modeling approach is proposed. To compensate the nonlinear, stochastic and time-varying uncertainties of the thermal drift, a probabilistic modeling approach is proposed. This project will provide the systematic modeling theory and approach for the accurate and fast positioning and control of the micro-manipulation robot. The theory and approach will be applied on the typical micro-manipulation robots.

微纳技术的快速发展将机电、制造、材料、生物和医学等许多领域的研究推进到微纳量级。微操作机器人是微纳技术在各领域中广泛应用的关键装备。为观察和操控微观目标，建立微操作机器人准确的系统模型对其精确快速的定位和控制至关重要。然而，由于系统中存在的各种动态不确定性，使得基于机理的建模非常困难，目前这类系统的建模还没有成熟的方法。本课题针对微操作机器人系统中的各种动态不确定性，通过对特征深入分析，结合先验系统知识，提出基于数据驱动的不确定性建模与补偿方法。针对微悬臂非线性、时空耦合和参数不确定性问题，提出微悬臂智能时空建模方法。针对压电驱动器蠕变、振动和磁滞多特征耦合不确定性问题，提出压电驱动多特征集成建模方法。针对热漂移非线性、随机和时变不确定性问题，提出热漂移概率建模方法。本研究将为微操作机器人系统的精确快速定位和控制提供系统化的建模理论和方法，结合典型的微操作机器人，形成理论和方法的应用技术。

微操作机器人是微纳技术在各领域中广泛应用的关键装备。为观察和操控微观目标，建立微操作机器人准确的系统模型对其精确快速的定位和控制至关重要。本课题针对微操作机器人系统中的各种动态不确定性， 提出基于数据驱动的不确定性建模与补偿方法。针对微悬臂非线性、时空耦合和参数不确定性问题，提出微悬臂智能时空建模方法。针对压电驱动器蠕变、振动和磁滞多特征耦合不确定性问题，提出压电驱动多特征集成建模方法。针对漂移非线性随机不确定性问题，提出漂移概率建模方法。本研究为微操作机器人系统的精确快速定位和控制提供系统化的建模理论和方法。