Method for solving nonlinear control problems using humans' abilities in manual control

人类可以在没有足够的被控对象信息的情况下完成各种复杂的控制任务。为了将人类的这种能力应用于广泛的非线性控制问题，我们提出了一种方法，以获得最佳的控制性能与手动控制的实验对应的真实的非线性控制任务。然后，我们从不同的角度来研究它的有效性。本研究的主要成果概括如下：1.通过改变条件，如显示风格，输入输出增益或时间尺度，良好的任务和它们的相关性的参数的范围进行了检查。它被证实，人的能力有效地为一定的一组参数. 2.在任务完成后，将非线性目标函数的得分反馈给操作者，操作者可以得到几乎最优的控制。这是一个重要的结果，人类可以修改他们的控制性能，甚至非线性指标.在挖掘控制的动力铲和位置控制的两个连杆臂，我们证实，操作员可以产生各种控制模式，导致良好的任务取决于目标函数。这些控制性能，这是无法获得的分析，是通过人类的能力。

Human beings can perform various complicated control tasks without enough information about controlled objects. In order to apply such humans' abilities to wide class of nonlinear control problems, we propose a method for obtaining best control performance with manual control for the experiment corresponding to the real nonlinear control task. Then we investigate its effectiveness from various viewpoints. The main results of this research are summarized as follows :1. By changing the conditions such as a display style, an input-output gain or a time scale, the ranges of the parameters for good tasks and their correlations were examined. It was confirmed that the human's ability works effectively for a certain set of parameters.2. Displaying the score of the nonlinear object function to the operator after task, the operator could get to perform almost optimal control. It is an important result that the human beings can modify their control performance even for nonlinear criteria.3. In the digging control of a power shovel and the position control of a two-link arm, we confirmed that the operator could generate various control patterns which lead to good tasks depending on the object functions. These control performance, which cannot be obtained analytically, were achieved by humans' abilities.