何毓琦的个人博客分享 http://blog.sciencenet.cn/u/何毓琦 哈佛(1961-2001) 清华(2001-date)


Is Control Dead? (II) 精选

已有 19924 次阅读 2015-5-4 11:04 |系统分类:海外观察

For new reader and those who request 好友请求,please read my 公告栏 first)

My 2010  article "Control is Dead?" http://blog.sciencenet.cn/blog-1565-344686.html produced a large amount of comments over the years. Although the phrase " control is dead" is  attributed to US NSF program officer in charge of research in systems, my article has

often been quoted without the "?" which gives a different flavor to casual readers. The latest response to my article http://blog.sciencenet.cn/blog-669170-880122.html  控制不死 by blogger 谢力,is but the most recent example. (note added 5/9/2015: On reflection, I realize the word "dead" was used origuinally for shock value. It arouses visceral and emotional reactions beyond what the substance of the debate here which is really "IS CONTROL A MATURE DISCIPLINE?")
The purpose this article is NOT to debate the validity of the title " Control is Dead" with or without the. "?" , but to supply some more recent development which can shed further light on the subject.  Readers can make up their own mind on what is meant by the phrase and decide on their own as to what to do in the spirit of my advices in the  original  2010  article.
An added fact (in 2013?)
The U.S. National Academy of Engineering voted to remove the word "control" from the list of disciplines the Academy considers as field for nomination. Despite protest from "control" members, this decision stood. Since then, the only new control member admitted to NAS membership was Richard Murray of Caltech (to my knowledge). And his qualification was based mostly on sophisticated applications rather than new theoretical developments. In the view,of the NAE, "control" like the tool "calculus or functional analysis" while enormously useful and prevails in all subjects, does not merit nomination of new members whose only contribution is in new third generation theoretical contributions. This view is by no mean unprecedented. Shannon Information theory was invented in 1948, it was enormously popular in the 50s and 60s. All the "Big Ox 大牛" at that time including Shannon himself were at MIT and member ship in NAE in the 80s were a large number of information theorists. However, this comes to an end in the 90s. The term "information theory" also disappeared from the NAE lexicon. Cryptography and communication theory became the replacements. I am not aware of pure Shannon information theorist qualified for NAE membership recently. Similarly, new control theory came of age 15 years later in the early 60s and is following a similar path in this early 21st century.
On the other hand, the subject of "Operations Research" being always anchored in and associated with Industrial Engineering and often mentioned in one breadth as IEOR has a small but application oriented section within NAE (section 8). It survives and this year even elected a new theoretical Chinese American member, Professor David Yao of Columbia http://blog.sciencenet.cn/blog-1565-865674.html . These fact explains somewhat the politics and popularity of scientific subjects, their  administration and reward system.  It can be argued that "control" is the victim of its own successes.
However, we are not here to merely review history but to also answer the question "what next" for  the younger workers in our field. If the answer to "control is dead?" or more  correctly "is control dead?" Is "not yet" then what is the second coming? Bioinformatics and some other new exotic "xxxcontrol", such as "quantum control" ,  are often mentioned as counter part of cryptography and communication network as new revival of information theory. In this respect, may I offer one word of caution to younger workers before they rush headlong into a topic such as "biologically inspired control theory". When the aerospace/control revolution began in the late 50s and early 60s, Newton's laws were already well understood for hundred of years. We did not have to learn anything new to write new theory and/or real world success stories. But in biology and possibly other subjects, we have a steep learning hill to climb before we can realistically make worthwhile contributions. In fact, our knowledge of biology are incomplete and not well understood, new discoveries may have to be made before we can effective "control". Beside getting credit or sharing credit as an outsider in an established field not of your own  are not always easy or fair. Basically, you have a make a serious  commitment of your career life in years in an uncertain environment before receiving any  recognition. For an already established scientist this may be fine. For a younger worker, this is a  risky endeavor that may not pay off. My personal recommendation in this respect is as follows:
1. Find an engineering problem that is ready made waiting for solution, such as, traffic, defense  against terrorism, smart xxx (whether it is buildings, car, et al), where the learning hill is easy to  climb or at least you are not at a disadvantage compared to others. Here also take advantage of the convergence of three forces - big data, sensor and GPS technology, and available decision, control, and computational tools.
2. Tried to use system control techniques to solve this problem
3. If you find that you have invent something new beyond current system control tools to solve the problem so much the better. You are well on your way to become famous.
This approach to re-invent control is totally consistent with my own career advice  that has served myself well and that I have advocated in previous blog articles. Two recent example are terrorism defense  and smart traffic management that are real world implementations and award winning. http://blog.sciencenet.cn/blog-1565-629054.html, http://www.nctspm.gatech.edu/content/nash-stackelberg-games-transportation-networks-leveraging-power-smartphones-traffic

Good luck to all system control researchers!


上一篇:[转载]Five Ways to Kill Your Dreams
下一篇:The 5th workshop on emerging topics in systems and control

31 马磊 谢力 程代展 沈律 朱豫才 李天成 康建 王善勇 唐常杰 胡俊 张文增 古槿 程娟 王洪涛 邵鹏 黄光伟 蔡宁 吴开宁 李贤伟 彭彬 赵凤光 王小平 苏金亚 王靖戈 icgwang xuexiyanjiu chenansb yunmu LongLeeLu zhangling qzw

该博文允许注册用户评论 请点击登录 评论 (36 个评论)


Archiver|手机版|科学网 ( 京ICP备14006957 )

GMT+8, 2019-3-23 22:40

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社