Hint
Example code: https://github.com/cselab/korali/tree/master/examples/learning/reinforcement/lander/
Lunar LanderΒΆ
A problem where a spaceship is to be landed on a surface as quickly as possible, yet with a softest possible contact. An example of the full game can be found here: http://moonlander.seb.ly/
In this case, the only the x,y position is passed as state. As a consequence, a recurrent NN is necessary to infer the velocity of the lander.