Car Racing#


This environment is part of the Box2D environments. Please read that page first for general information.

Action Space

Box([-1. 0. 0.], 1.0, (3,), float32)

Observation Shape

(96, 96, 3)

Observation High


Observation Low





The easiest control task to learn from pixels - a top-down racing environment. The generated track is random every episode.

Some indicators are shown at the bottom of the window along with the state RGB buffer. From left to right: true speed, four ABS sensors, steering wheel position, and gyroscope. To play yourself (it’s rather fast for humans), type:

python gym/envs/box2d/

Remember: it’s a powerful rear-wheel drive car - don’t press the accelerator and turn at the same time.

Action Space#

There are 3 actions: steering (-1 is full left, +1 is full right), gas, and breaking.

Observation Space#

State consists of 96x96 pixels.


The reward is -0.1 every frame and +1000/N for every track tile visited, where N is the total number of tiles visited in the track. For example, if you have finished in 732 frames, your reward is 1000 - 0.1*732 = 926.8 points.

Starting State#

The car starts at rest in the center of the road.

Episode Termination#

The episode finishes when all of the tiles are visited. The car can also go outside of the playfield - that is, far off the track, in which case it will receive -100 reward and die.


lap_complete_percent dictates the percentage of tiles that must be visited by the agent before a lap is considered complete.

Passing domain_randomize=True enables the domain randomized variant of the environment. In this scenario, the background and track colours are different on every reset.

Passing continuous=False converts the environment to use discrete action space. The discrete action space has 5 actions: [do nothing, left, right, gas, brake].

Version History#

  • v1: Change track completion logic and add domain randomization (0.24.0)

  • v0: Original version


  • Chris Campbell (2014),


Created by Oleg Klimov