💻 Unity 2024 ML-Agents | Live AI Robot Training | Kyle 2a0a | PyTorch | Part 9
2In this video, I will show you how to train a robotic walker to cooperate with other walkers in a hostile environment using Unity Machine Learning Agents Toolkit.
**Coding** up the remaining part of the Motion Perception System which takes in a huge array of ray's and what they hit, and somehow transforms that data into a direction and speed which the agent should take. This also uses a focus sphere to determine the direction
ML-Agents is an open source project that allows you to create and train intelligent agents using reinforcement learning and deep neural networks. In the previous episodes, we added different features to our robotic walker, such as a negative reward for colliding with another walker, a hostile walker that can attack them, and a reward function that encourages cooperation and survival.
In this episode, we will add a new twist to our robotic walker game: we will remove the stabilizers that keep the ragdoll upright. This means that the ragdoll will fall on the ground and have to learn how to get up by itself. We will see how this affects the training process and the performance of our robotic walker. We will also compare the results with the same walkers that still have the stabilizers.
By the end of this video, you will have a deeper understanding of how to use ML-Agents to create and train agents with complex and realistic physics. You will also have a more challenging and fun robotic walker game with more emergent and diverse behaviors.
This is the seventh video in my Unity Machine Learning series, where I teach you how to use ML-Agents for various games and simulations. If you want to learn more about ML-Agents, check out my other videos and playlists on this topic.
If you enjoyed this video, please hit the like button and subscribe to my channel for more Unity Machine Learning content. And don’t forget to leave a comment below and let me know what you think of my robotic walker game. Thanks for watching!
#unity #gamedev #ai #mlagents #robot #walker #physics #2023 #visualstudio