The ego-vehicle is driving on a multilane highway populated with other vehicles. The agent’s objective is to reach a high speed while avoiding collisions with neighboring vehicles. Driving on the right side of the road is also rewarded. I am using OpenAI gym. It is a toolkit for developing and comparing reinforcement learning algorithms.
I am using gym version 0.21.1. Follow the instructions here for the installation of highway environment.
For the lane changing task, I am using discrete meta action:
0 : lane left
1 : idle
2 : lane right
3 : faster
4 : slower
Refer the doc for more details.
The GrayscaleObservation is a W x H grayscale image of the scene, where W, H are set with the observation_shape parameter. The RGB to grayscale conversion is a weighted sum, configured by the weights parameter. Several images can be stacked with the stack_size parameter, as is customary with image observations. Refer the doc for more details.
