this project is for generating a model of an autonomous car in a Udacity's car simulation using convolutional neural networks
I started this project to learn more about automation, simulation and neural networks
for this version you need to install the udacity/self-driving-sim it is also recommended that you install version 2 since it is the version that I have used for this project
it is also very important that you have all the required libraries and that you install the simulation in the same directory as the code file
after extracting the simulation in the same file as the project create an empty file which you should name (Simulation_data) then run the simulation and select the Training mode then select the (Simulation_data) folder then run the training.py file notice that this will take some time as it trains your model afterwards you'll see a file in the same directory by the name (Alastor.h5) this is your model so you only have to run the training.py once then open the simulation and select the map and select autonomous mode then run the Test_model.py file which should run the simulation autonomously
this model is based on Nvidia's End-to-End Deep Learning for Self-Driving Cars which you can check from here
this was also inspired by Murtaza Hassan's project on a similar subject
I tested it in 2 maps to see how well the model generalized
Apparently it did generalize well although it wasn't as good as it was in the initial map
this project can only work on Udacity's simulation and cannot be used for real Autonomous Cars
I have started working on the CARLA Simulator but it might take some time to finish because of all the exams and stuff
check this out if you're looking for the CARLA Simulation Project
The next version will use CARLA Simulator and will have a more updated and specific architecture for the model that it generates which will hopefully produce better predictions and results
Due to some technical and personal issues the project will be delayed for a while