I'm working on a program that will find the fastest path from one point to another (think mapquest or a Garmen gps) using a genetic algorithm. Its been a while since I've worked on genetic algorithms, and they were all used in tandem with neural networks, so I'm anxious to see what the results will be like. The program will load a map, ask the user for beginging and end points, calculate the best route, then display the path.
Genetic Algorithms work as such:
- Generate a random set of commands. This will be one 'organism'.
- Generate a bunch of other 'organisms' with different commands.
- Run each set of commands through the map.
- Record the results and score the 'organism' on its 'fitness level'.
- Generate a new set of organisms based on how well the previous ones did.
- Repeat steps 3-5 several times.
