Direct Design of Human Skill and Its Implementation on Robot Using Model-free Controller Synthesis

本研究在不建立任何数学模型的情况下，提出了基于人体操作数据的控制器整定方法。由于没有建立数学模型，在控制器实现之前，我们不知道所设计的控制器是否真正稳定了被控对象。首先，作为基础研究，提出了基于被控对象输入输出数据的闭环系统稳定性评价方法，并将其融入到控制器整定的非线性和线性优化问题中。此外，针对线性优化问题，提出了保证闭环系统稳定的控制器参数整定方法。

This research proposed the controller tuning method from human operation data without constructing any mathematical model. Since the mathematical model is not constructed, we have no idea in advance of the controller implementation whether the designed controller actually stabilizes the controlled plant or not. Firstly, as a basic study, the evaluation method of the closed-loop stability from the input/output data of the controlled plant was proposed, and it was incorporated into the nonlinear and the linear optimization problem for the controller tuning. Moreover, in the case of the linear optimization problem, the controller parameter tuning method guaranteeing the closed-loop stability was proposed.