bruno147 / fsmdriver

A finite state machine controller for TORCS environment.
3 stars 4 forks source link

Could you give some more details about your genetic optimization method? #10

Closed FinalTheory closed 6 years ago

bruno147 commented 8 years ago

The FSMDriver uses 17 racing parameters (4 for stuck, 7 for outOfTrack and 6 for insideTrack). The parameters are float point numbers and their binary representation was used in order to create an individual. An individual has a total of 17 parameters each of them represented by 32 bits resulting in a 544 bits string. The GA used standard configurations from literature, the population size was arbitrarily set as 30, which leads to 616 generations for a moderate problem complexity. Selection is done through elitism by keeping the best 4 individuals from a generation unchanged, and the other 26 are generated from the best 10 individuals’ reproduction. This step is done by randomly choosing a pair from the top 10 that are cloned with a 5% chance or generate a pair of offspring with a 95% probability. These offspring are produced through crossover by partitioning the parents’ bitstrings at a random position and swapping them. Finally, mutation (a simple bit flip) is applied to every bit in every individual of these 26 with a 1% probability.

Gastd commented 6 years ago

More information can be seen in our paper