教我们飞行力学的孙教授，他的研究方向是飞行器控制，我对他的研究很感兴趣

教我们飞行力学的孙教授，他的研究方向是飞行器控制，我对他的研究很感兴趣，所以加入了他的实验室。
第一次参与科学研究是在大学三年级，我和两位队友组成团队，在孙教授的指导下，参加了学校的科研训练。
我们的项目是一个关于伪谱法得最优控制器，研究时间是十个月。这个控制器运行在嵌入式平台上，用于计算飞行器的飞行轨迹。

研究工作刚开始的时候进行的很慢，因为我们都是第一次研究大型的课题，我们得基础知识不够，而且还有课程表里得其他课程要学习，
所以压力很大。我们学习了数值方法，非线性规划，最优控制原理等基础知识，然后开始阅读相关得文献，了解最新的
研究结果，这些工作花费了四个月的时间。

伪谱法是一种求解非线性最优控制问题的数值算法，是最优控制领域的研究热点。我们的任务是，按照最新的研究结果实现伪谱法。

考虑到项目对快速性的要求，我们选择了C++作为编程语言，利用开源的矩阵库Armadillo作为基础的线性代数库，
参考了开源的最优控制软件gpops，psopt。我们对算法核心进行了改进，使用了最新的RPM方法，并加入了自适应配点算法。

代码有10000多行，重写了两次，最终用了三个月写完。最后得测试结果比现有得matlab版本的实现要快十倍，满足项目要求。在研究过程中，
我们参考和使用了大量的开源软件包，因此我们决定把自己的软件也开源，发布在开源网站上，分享给其他的研究者。

整个研究过程很累，但是我觉得非常值得，因为学到了很多东西。经过自己实践以后，我对微积分、线性代数和最优控制理论有了更深刻的理解。
知道了怎么研究一个实际的问题，以及如何与其他人合作研究。在研究过程中进行学习，知识掌握得最快最好，因为，有一点不懂的话，整个项目就会失败。

結果 (中文) 3:

Professor Sun, Professor Sun of our flight mechanics. His research interests are aircraft control, and I am very interested in his research.The first time involved in scientific research is in the third year of college, I and two team members of the team, under the guidance of Professor Sun, participated in the school's research and training.Our project is a pseudo spectral method of optimal controller, study time is ten months. The controller runs on the embedded platform, which is used to calculate the flight path of the aircraft.At the beginning of the research work is very slow, because we are the first large-scale study of the subject, we have the basic knowledge is not enough, and there are other courses in the curriculum to learn,So the pressure is great. We learned the basic knowledge of numerical methods, nonlinear programming, optimal control theory, and then began to read the relevant literature, to understand the latestThe results of the study took four months.Pseudo spectral method is a numerical algorithm for solving nonlinear optimal control problem, which is a hot research topic in the field of optimal control. Our task is to implement the pseudo spectral method according to the latest research results.Taking into account the requirements of the project, we choose C++ as the programming language, using the open source matrix library Armadillo as the basis of the linear algebra library,Reference to the source of the optimal control software gpops, psopt. We improved the core of the algorithm, using the latest RPM method, and joined the adaptive algorithm.Code has more than 10 thousand lines, rewritten two times, and ultimately spent three months to finish. The final result of the test results than the existing matlab version of the implementation to be nearly ten times faster, to meet the project requirements. In the course of the study,We refer to and use a large number of open source software packages, so we decided to put their own software is also open source, released on the open source website, to share with other researchers.The whole research process is very tired, but I think it is very worthwhile, because I learned a lot of things. After my own practice, I have a deeper understanding of calculus, linear algebra and optimal control theory.Know how to study a practical problem, and how to cooperate with other people. In the process of learning to learn, knowledge is the best, the best, because, if you do not understand, the whole project will fail.