Clarification about ML Agent

[RanjaniN-10]

Hi, I am new to this project.I would like to know did you use ML Agents of unity in this project. If so, can you please guide me with the tutorials/links to learn more about it. I would like to recreate this for my college project on my own can you please guide me.

With curiosity, Ranjani N

[Kyushik]

Hello. Thanks for using my simulator! I used Unity ML-Agents in this project for making reinforcement learning environment. I think there is tutorial in the ML-Agents official github I wrote a book about the tutorial of the Unity ML-Agents, but it is published in Korean. If you want to fix the environment, you should know how to use the Unity. If you are not familiar with Unity, there is good tutorial in the Unity official webpage

[RanjaniN-10]

Hi,Thanks for your attention. Can you please recommend me some Unity ML-Agents books in English (or You tube tutorials).

[Kyushik]

I just studied ML-Agents using the examples and github, so I don't know any other book or youtube tutorials.

[RanjaniN-10]

Thanks for your reply. I'll start learn from here https://github.com/Unity-Technologies/ml-agents/blob/master/docs/Learning-Environment-Create-New.md

[RanjaniN-10]

Hi @Kyushik System Details: OS: Ubuntu 18.04 Processor: Intel® Core™ i3-8130U CPU @ 2.20GHz × 4 Memory: 4 GB

I have inserted all the three assets (Simple Racer, Simple Town, Simple Sky) in Unity_SDK and installed the ml-agent (0.6 version).

I still found the Error (Please check the screenshot) Also the Agents run only in the first lane. Could you Please help me to solve this.

Thanks Ranjani N

[RanjaniN-10]

Hello @Kyushik I am a beginner in Tensorflow. I received an error regarding TFsession and TFgraph(please check the screenshot). Could you please help me with this error.

[Kyushik]

At the bottom of the video, it says "The brain was set to inference mode but the Tensorflowlibrary is not present in the Unity Project" Therefore, you should connect player brain to the Agent. You need to check the brain of the Agent and if it is "VehicleBrain", You should change it to Player brain. Then you can control the vehicle with your keyboard. If you want to train the agent with reinforcement learning algorithm. You should check the "Control" which is next to the VehicleBrain in Academy and build the environment.

[RanjaniN-10]

Hello @Kyushik Thanks a lot for your reply.I have successfully created executable file. Then I tried to train it with your RL Algorithms(DQN_sensor.ipynb). While executing examine the observation and state spaces I got the following error. IndexError: index 0 is out of bounds for axis 0 with size 0

[myxyzy]

Hello Kyushik, Is it possible to pull reward function(values) you defined in unity to python code. I would like to modify rewards and implement other algorithms.

[Kyushik]

@mustafayildirim16 Hello! You can change the reward function in the Unity. You should change vehicle agent code!