In order to reach our goals, we preferred an experimental approach. Simulation is not appropriate especially because of the visual input. Visual inputs are hard to simulate properly. Even if we can simulate proper visual inputs by using physics of light for instance, it will require a lot of computation time.
Furthermore, the real world is such complex that is almost impossible to model it correctly. The simulation will suffer from this lack of complexity and, when tested, the robot will perform badly. Moreover, it is also hard to simulate noise. Unfortunately, noise is often a key issue in robotics. Thresholds are often adjusted according to the amount of noise. Therefore it is better to make real experiments instead of simulations.
For this project, experiments were made without using simulation.