Sunday, October 21, 2007

Evolving Missile Trajectory Through Simulated Evolution

I have found an old school evolution strategy article, from 1980: Optimization of missile trajectories by means of evolution strategy. You can read the first page here.

From the abstract:

Evolution Strategy is an optimum seeking method which attempts to apply the rules of biological evolution - mutation and selection - as closely as possible to technical optimization problems.
In the paper results are presented showing the successful application of evolution strategy to missile trajectory optimization problems like

- Range optimization of a ballistic rocket and e boost-glide missile

- Optimization of miss distance due to lateral wind of a Short range antitank missile

- Optimal trajectory shaping A comparison with other optimization methods is given to show the efficiency of evolution strategy.


From the introduction we can understand a little of the academic Zeitgeist along that period, see below:

"The paper deals with solutions of optimization problems which we have gained by the application of evolution strategy to missile engineering and development. Evolution strategy is an imitation of nature's way of optimization, a game of mutation and selection. Since nature has been man's teacher in many ways it might appear quite obvious to apply the laws of biological evolution to problems of engineering and technology.

On the Other hand there are also objections to that method, the most important ones being that "cut and try" is a sign of poor engineer and craftsmanship and that we cannot afford nature's lavishness in means and in time to wait for the result. Based on a number of number of encouraging results [1], [2] we have tried to gather our own experience with the matter and will present the results we have achieved.


It seems that the author worked to the famous German aircraft manufacturer Messerschmitt, which gave birth to the world's first operational turbojet fighter aircraft, the so-called Messerschmitt Me 262.

Below the old Messerschmitt logo.

Labels: , , , , , , , ,


Anonymous Anselmo Pitombeira said...

The author has claimed that craftsmanship is poor engineering. Maybe at the time (beginning of the eighties) there still was a general feeling among scientists that we could solve all problems by means of analysis. Nevertheless, we face even more complex problems that our mathematical models have failed to solve. Engineering always was a matter of trial and error, so that successful results are documented and make part of the Engineering body of knowledge.

P.S. I couldn't visualize the confirmation image in Firefox, only i IE (irgh..) Is that a problem with blogger site?

22 October, 2007 10:07  
Blogger Nosophorus said...

Hi, Anselmo! :)

Thank you for your comment!

From what I could understand, the author's claims walk toward the opposite direction. He said that, despite all the benefit outcomes the method had given until that moment, there were (and, in my opinion, still there are!) objections concerning the method he applied (that is, the evolution strategy). He did not claim that craftmanship is poor engineering, on the contrary, when someone decides to use an evolutionary algorithm (as the author did), the craftmanship, simplicity, and cleverness are very important (for instance, to design an well suited fitness/objective function, to choose a reasonable solution representation, which evolutionary operators apply, how robust the EA can be, and etc.).

Upon the problem solving through analysis, frequently an analytical model is not available or the problem by itself cannot be tractable via analysis, let alone the possibility that, even though there is an analytical model, handling it can be very expensive (computationaly, monetarily, etc) or even impossible (taking into account the current tools we have). Take, as an example, the Two Phase Flashing Nozzle Experiment made by Professor Hans-Paul Schwefel in the 1970s. At that time, and nowadays, there is not a simulation model related to that problem and the experiment was performed without computer! Even today, it is not possile to calculate what happens within such nozzle: thermodynamically far away from equilibrium, drag between slow water droplets and fast steam, three-dimensional turbulent boundary layer with liquid sublayer, supersonic behind nozzle throat, etc. So, when dealing with problems like that, the craftmanship must arise, since feasible results depend on it! :)

Upon the browser you mentioned, you should check if all plugins are installed. Blogger gives only the hyperlink to the document. :)

Até Mais!


22 October, 2007 17:22  

Post a Comment

Links to this post:

Create a Link

<< Home

Charles Darwin Has A Posse Check Google Page Rank