What you learned

This learning path showed how ML Agents run well on Arm devices and give us great game experiences. It also showed how to:

  • Implement game AI using machine learning
  • Train machine learning agents to provide varying skill levels

Next steps

Later learning paths will address deploying to an Arm-based Android device and how to use the Unity Profiler tool to analyse performance.

Stretch tasks

If you would like to take this learning path further, try the following:

  • Deploy to a Windows Arm device using the build option for UWP (Universal Windows Platform)
  • Teams (multiple enemies and/or co-operative with the player) - you’ll need to train them to work together. Hint: look at adding team rewards and possibly self-sacrifice (so one character may sacrifice itself for the betterment of the team. The team reward would have to make such an action worthwhile). Also look at training the team characters together.
  • Non-battle characters for different sorts of characters, games and different conditions for winning and losing.

Further reading

If you are interested in exploring the Unity ML Agents toolkit or the subjects referenced in this learning path beyond what we have covered here, you can follow these links: