CarRacing | learn unreal engine | Game Engine library

Windows support is at present moment experimental (source). That being said, on most of the occasions you will get it to work, but some of the functionality could be broken. Proceed with caution.

Here's what could work:

1. Install Anaconda.
2. Start Anaconda command line (it adds conda to the PATH).
3. conda create -c conda-forge -n gymenv swig pip - create base environment.
4. conda activate gymenv - activate it.
5. pip install Box2D gym - install base gym with Box2D.
6. (Optional) pip install gym[all] - install all components of Gym. You might run into errors.

If you get build errors on missing compilers, install Visual Studio Built Tools.

Depending on the component you're using, you might need e.g. pystan. Activate then the conda environment and install it, i.e. conda install pystan.

An alternative could be to try to install Gym directly through conda:

I am not using anaconda but virtualenv. Nevertheless, I had initially the same error. It seems that the "Classical" Environments were working but not the Box2D games. I tried to install Box2D separately with "pip install Box2D" but I received errors for a missing swig.exe when installing dependency "wheel".

To solve this,

• I downloaded swig and then added the path to the swig.exe to my windows path.
• Then I uninstalled gym and Box2D.
• Then I installed gym and then Box2D with "pip install gym[box2d]"

You use self.initialize() in while-loop but self. can be used only inside class

You should rather use running = True/False to exit while loop and later use car_racing.initialize()

Box means that you are dealing with real valued quantities.

The first array np.array([-1,0,0] are the lowest accepted values, and the second np.array([+1,+1,+1]) are the highest accepted values. In this case (using the comment) we see that we have 3 available actions:

1. Steering: Real valued in [-1, 1]
2. Gas: Real valued in [0, 1]
3. Brake: Real valued in [0, 1]

Given that there was not answer I have modified the environment by reading and understanding the code, it is not that hard, I will leave the repo just in case it is useful for someone else

https://github.com/NotAnyMike/gym

Some screenshots of the modifications:

Complex Map:

Complex intersections:

And some obstacles:

To see more and the updated version see the repo on github.