prfernando
commented
11 years ago The failures are implemented by the EA - hence before simulation.
Do you mean you want each fitness evaluation of the individual to have a different starting position? This is not possible with the current version of the code. Unfortunately we do not have the time to implement this now : (
A hack would be to modify the evaluator code to write the name of a new experiment config file for each evaluation - which in turn would contain a different startPosition filename. Each startPosition file will contain one line with the desired start position.
Our group is doing robustness-chasing in which we want to evaluate each individual 3 times for statistical purposes. We also want to have 3 starting positions.
Are the failures (of actuators and sensors) done before the starting position or after in calculating the fitness? I was wondering because it could help us cut down our already lengthy runtime.