Banana Collectors

A copy of the Unity ML-Agents' Banana Collector environment. The agents are trained to collect bananas, shoot at each other and avoid the bad bananas. Discrete action branching is used in this example.

This example shows how to create an environment where the action space is discrete with action branching, the agents can interact with each other. and how to use raycasting results as the agent’s observations.

We copied the environment from Unity ml-agents, and modified it to train in editor. Here are the results:

Before Training

The movement of agents are random before training

After Training

Agents can shoot at each other and collect the good bananas

Go to Sourcecode

This Banana Collectors example is one of the examples in the the UnityTensorflowKeras repository. Go to the repository from the link below to install it according to the instructions.

The Banana Collectors example is located under Assets/UnityTensorflow/Examples/BananaCollectors directory.

For more information about this example, see Here.

Exercises

NA