Reinforcement Learning for autonomous driving Python code walkthrough

preview_player
Показать описание
In this video we go through Python code behind a very basic way to get Reinforcement Learning working with Carla simulator.
A high level concept overview, please refer to a previous video here:
My GitHub with the code:
Рекомендации по теме
Комментарии
Автор

Hi, I'm doing a bsc project on carla and rl your tutorials have been a great help!

TrinitySx
Автор

Fantastic! I'm very excited to get started!!

MakeDataUseful
Автор

HI, I would like to have some more information regarding the training of the CNN for feature extraction. Is it only trained on images from CARLA segmentation? How many images is it trained on? How were they acquired? Thank you for your attention.

matteomemoli
Автор

do i have to install carla and unreal engine for this (it takes disk space of 165gb)
please reply

peddikarthik
Автор

Hello Could you please share requirements ? Iam trying to run this but i think i dont have installed correct version of tensorflow for example as only import is not working for me
Import "tensorflow.keras.models" could not be resolved
Thank you in advance :)

romanpires
Автор

Amazing job! With this "train" code the car should be able to connect intersections correctly? Thanks for this serie of videos

danielc
Автор

I do not understand the concept of steering lock, could you explain it for me?

rowanabdelaleem
Автор

This is a nice tutorial. i already have my pytorch saved model and would like to test it in the carla environment, how would that work for me.

chinny-qu
Автор

How do I know when to stop it? My "model" folder is not filling up with files, only empty folders. When does the algorithm stop? I want to see the data but I am not able to, please help

drumzstringz
Автор

Hi, great work and walkthrough! Is there a particular reason you chose to train and deploy your own CNN model to parse the image information instead of using the SB3 PPO predefined CnnPolicy instead of the MLP one? Just curious

pigo
Автор

Thanks again! This is incredibly helpful.
I have a question can we implement transformers (attention mechanism) with the PPO algorithm?

nauman_
Автор

Hey sir I appreciate your work, My final year project is Autonomous car in Carla, so to be sure have you reached to a desirably working model, I mean car stops in red light, avoid collisions and can go from point A to point B, if yes or not can u post the last footage of the model, thanks a lot, I would be so happy if you answered me soon

Neyşeker
Автор

Hi, can you please tell me what I need to change in the code to run training on a GPU?

АндрейСеменов-чх
Автор

Hello, these tutorials have really been helpful.
I have been trying to train the model for steering but it always stops half way because the random spawning causes this error;
Traceback (most recent call last): File "c:\Users\user\Desktop\SelfDrive\SelfDrive-master\RL_Full_Tutorial\environment.py", line 214, in reset self.vehicle = self.world.spawn_actor(self.mitsubishi, self.transform) RuntimeError: Spawn failed because of collision at spawn position

How do I solve this?

atuhairedruscilla
Автор

Hello sir, I am getting this error: in from gymnasium.wrappers.monitoring import video_recorder

No module named

sonas
welcome to shbcf.ru