有哪些機器人、無人機、控制等「軟+硬」領域大牛寫的非技術類作品？
包括書籍、傳記、科普、人生感悟、博客、專欄等
首推卡耐基梅隆大學金出武雄教授的《像外行一樣思考，像專家一樣實踐——科研成功之道》，此書是金出教授長期在日本和美國生活和科研的人生感悟。博士畢業後看的這本書，非常有共鳴感，覺得句句說到心窩裡。金出武雄觀察到的日本學者在寫作、演講和科研方法等方面的問題，對中國人也是適用的。比如，和歐美人說話時，不習慣看著對方眼睛；演講時不是努力講個引人入勝的好故事，而是在背誦演講稿等等。
此書原始版本是日語：素人のように考え、玄人として実行する―問題解決のメタ技術沒有英文版，中文版翻譯水平非常差。可能沒有類似世界一流實驗室苦逼科研經歷的人不太明白書中所述道理，此書在豆瓣上評價並不高。
http://book.douban.com/subject/1867455/
再，Ivan Sutherland 的 《技術與勇氣》（Technology and Courage）
http://cseweb.ucsd.edu/~wgg/smli_ps-1.pdfSunLab整個Perspective Essay Series的幾篇文章都非常值得一讀。Richard Hamming``You and Your Research""，這個有非常好的中譯版本
http://www.rangang.com/archives/9Hamming晚年所著「Art of Doing Science and Engineering: Learning to Learn」
後面兩個是公認值得讀了再讀的經典之作，覺得不好絕對是讀者本人水平不夠。
最近發現的一本：諾貝爾經濟學獎得主 Herbert Simon 的 《Model of My Life》。Simon同時還是政治學家，社會學家，心理學家，計算機科學家（人工智慧的開山鼻祖）。
UIUC 計算機系謝濤教授的 Advise 系列文章，對中國學生學者尤其有針對性：
Advice for Researchers and Students by Tao Xie斯坦福Philip Guo（計算機博士）的 「The Ph.D. Grind」Philip Guo - The Ph.D. Grind展示真實的名校 Ph.D. 日常科研生活，以及學術界的「一將成名萬骨枯」。維納的《控制論——關於在動物和機器中控制和通訊的科學》《人有人的用處》
Illah的《機器人與未來》http://blog.exbot.net/archives/734金觀濤《控制論與科學方法論》錢學森《論系統工程》
溫伯格《系統設計的一般原理》其他回去加不算新的一個：
Control in an Information Rich World
- Report of the Panel on Future Directions inControl, Dynamics, and Systems
http://www.cds.caltech.edu/~murray/cdspanel/report/cdspanel-15aug02.pdf
相關：
Future directions in control in an information-rich world
IEEE Xplore AbstractMathematical Models: Uses and Limitations
IEEE Xplore Abstract
節選
Mathematical modeling is a technique that engineers use to solve any practical problem. Its severe limitations and pitfalls are illustrated.
In 1958, when someone commented to me that it was fortunate that the defense program lent itself so readily to mathematical formulations, I replied that this was largely illusory. In fact, I asserted, if the Federal government suddenly decided that the urgent national problem was garbage collecting, and funded it accordingly, there would soon be a mathematical theory of garbage collecting. But imagine my surprise, ten years later, to discover that garbage collecting has become an urgent national problem, under the more esoteric name of "waste management"-and, sure enough, mathematical modeling is being applied. So far, the funding level is only in hundreds of thousands of dollars, but as it gets up into the millions, there will no doubt be a differential theory of garbage collecting, a topological theory of garbage collecting, and a statistical theory of garbage collecting.
The principle of mathematical modeling has now become so widely accepted in the physical, biological, and that it scarcely seems necessary to present a case for it. Accordingly, I have prepared the accompanying chart (Table I) listing the pitfalls inherent in modeling. Perhaps one illustration provides the best overall advice to mathematical modelers; a rocky road, filled with potholes, and a large sign in the foreground, warning "Proceed-With Caution."
---No model is ever a perfect fit to reality. Deductions based on the model must be regarded with appropriate suspicion.
1. Don"t believe the 33rd-order consequences of a first-order model.
The "impenetrability" of the sound barrier.
2. Don"t extrapolate beyond the region of fit.
The world is flat, locally.
3. Don"t apply any model until you understand the simplifying assumptions on which it is based, and can test their applicability.
Perpetual motion inventors and angle trisectors have not read the fine print.
---Distinguish at all times between the model and the real world.
1. Don"t believe that the model is the reality.
Sticking pins in a voodoo doll.
2. Don"t distort reality to fit the model.
Gestalt perception of preconceived patterns (e.g., Martian canals?).
3. Don"t limit yourself to a single model. More than one may be useful for understanding different aspects of the same phenomenon.
Light is a wave or a particle, and heat and electricity can be regarded as fluids.
---
A model must be permitted to evolve as conditions change or as additional data become available.1. Don"t retain a discredited model.
Astrological forecasting. 2. Don"t fall in love with you model.Network theorists who do not wish to admit any of the newer components into their circuits.3. Don"t reject data which are in conflict with the model. Use them to refute, modify, or improve the model.Astronomers pooh-poohed meteorites for centuries. Hypnotism was not believable.推薦閱讀：