An implementation of an Autonomous Vehicle Agent in CARLA simulator, using TF-Agents

Python 100.00%

dqn dqn-tensorflow self-driving-car autonomous-vehicles av computer-vision simulation deep-reinforcement-learning carla-simulator vehicle

autonomous-vehicles-adaptive-cruise-control's Introduction

New version of Adaptive-Cruise Controller using DRL:

https://github.com/kochlisGit/Noise-Adaptive-Driving-Assistance-System

Self Driving Vehicles Using Deep Reinforcement Learning

** Description **

This is the project of my thesis. I've implemented an agent in CARLA Simulator, which is capable of navigating a vehicle safe & fast, using only 2 front cameras. More info about the simulator can be found here: https://carla.org/ . The agent has learnt to navigate in a lane using deep reinforcement learning algorithms. The development of the agent was made in Python.

Demonstration

This directory contains 8 video files (mp4) that demonstrate how the vehicle is moving in the simulator https://github.com/kochlisGit/autonomous-vehicles-agent/tree/main/videos

The validation of the vehicle was done in a pre-defined route, that was new to the agent. The results are astonishing!

Python Libraries

Carla API
Numpy
Matplotlib
Tensorflow
Keras
TF-Agents
Tensorflow-Addons

Sensors

Collision Detector: https://carla.readthedocs.io/en/latest/ref_sensors/#collision-detector
RGB Camera with Semantic Segmentation: https://carla.readthedocs.io/en/latest/ref_sensors/#semantic-segmentation-camera
Depth Camera: https://carla.readthedocs.io/en/latest/ref_sensors/#depth-camera

Algorithms:

DQN: https://www.cs.toronto.edu/~vmnih/docs/dqn.pdf
N-Step DQN: https://arxiv.org/pdf/1901.07510.pdf
DDQN: https://arxiv.org/pdf/1509.06461.pdf
C51: https://arxiv.org/pdf/1707.06887.pdf
Decaying Epsilon Greedy: http://tokic.com/www/tokicm/publikationen/papers/AdaptiveEpsilonGreedyExploration.pdf
Convolutional Neural Network CNN: https://iopscience.iop.org/article/10.1088/1742-6596/1004/1/012028/pdf
Yogi Optimizer: https://papers.nips.cc/paper/2018/file/90365351ccc7437a1309dc64e4db32a3-Paper.pdf
GELU: https://arxiv.org/pdf/1606.08415.pdf

Execution

First, You have to download carla and all the libraries above. Then, download my "code" directory and paste it into "Carla/PythonAPI/". Run "agent/straight_lane_agent_c51_training.py" to start the training.

IMPORTANT The simulator window should be open, in order for the training to occur. Check https://carla.readthedocs.io/en/latest/start_introduction/ for more information of how to setup carla.

Libraries

Python 3.7
Carla <= 0.12
Numpy >= 1.15
Matplotlib
Tensorflow >= 2.0
TF-Agents >= 0.13
Tensorflow-Addons >= 0.13 1 Opencv >= 4.0

autonomous-vehicles-adaptive-cruise-control's People

Contributors

Stargazers

Watchers

Forkers

bill-me lli7 studylad 12kobe gavin-tao akm108 honeytx

autonomous-vehicles-adaptive-cruise-control's Issues

python3 straight_lane_agent_c51_training.py

How to solve this problem?
(Autonomous-vehicle) ma@ma-ROG-Strix:~/Desktop/carla/PythonAPI/Autonomous-Vehicles-Adaptive-Cruise-Control/code/agent$ python3 straight_lane_agent_c51_training.py
2023-12-05 15:28:04.402854: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 AVX_VNNI FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2023-12-05 15:28:04.466888: I tensorflow/core/util/port.cc:104] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable TF_ENABLE_ONEDNN_OPTS=0.
2023-12-05 15:28:04.868101: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer.so.7'; dlerror: libnvinfer.so.7: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-11.2/lib64:/home/ma/carla-ros-bridge1/catkin_ws/devel/lib:/opt/ros/noetic/lib
2023-12-05 15:28:04.868145: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer_plugin.so.7'; dlerror: libnvinfer_plugin.so.7: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-11.2/lib64:/home/ma/carla-ros-bridge1/catkin_ws/devel/lib:/opt/ros/noetic/lib
2023-12-05 15:28:04.868149: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.
Traceback (most recent call last):
File "straight_lane_agent_c51_training.py", line 34, in
from environment.carla_environment import CarlaEnvironment
ModuleNotFoundError: No module named 'environment.carla_environment'; 'environment' is not a package

Running repo

Hi ,Hope you are well
Hi , I was trying to run this repo but I can’t , can you please tell me how to make testing after training and validation?
What Carla version fits this repo?
I have Carla 0.9.11 ,does it suit your model?
Can I use CPU instead of GPU?
can I run this repo on windows?
Sorry for making it long .
Thanks in advance

python code problem（straight_lane_agent_c51_training.py）

I have changed the path, but the execution still fails.
Please tell me the correct path.
try:
sys.path.append(glob.glob('../../carla/dist/carla-*%d.%d-%s.egg' % (
sys.version_info.major,
sys.version_info.minor,
'win-amd64' if os.name == 'nt' else 'linux-x86_64'))[0])
sys.path.append('/')
sys.path.append('../simulation2')
sys.path.append('../environment2')
sys.path.append('../../carla/')
Thanks.

How to obtain the data from the sensors??

i have found that at ur file simulation at simulation.py u defined a things like sensor_data and sensor_callback . so how do i get the readings of sensors u have implemented in ur code???

Hello , I don't know how to solve this error " File "C:\Users\MARYAM\Documents\po\lib\site-packages\tensorflow_addons\optimizers\yogi.py", line 101, in init self._set_hyper("learning_rate", learning_rate) AttributeError: 'Yogi' object has no attribute '_set_hyper'"

TypeError: No to_python (by-value) converter found for C++ type: class std::vector<unsigned char,class std::allocator<unsigned char> >

Hi, thank you for sharing your work! It's been really helpful.
Unfortunately I have this error when I run "python straight_lane_agent_c51_training.py".
lambda sensor_data: self._sensor_callback(SensorType.COLLISION_DETECTOR, sensor_data)
File "D:\Carla9.9\WindowsNoEditor\PythonAPI\code\agent\simulation\simulation.py", line 140, in _sensor_callback
data = sensor_data.other_actor.semantic_tags
TypeError: No to_python (by-value) converter found for C++ type: class std::vector<unsigned char,class std::allocator >

Can you please help me? Thank you.

kochlisgit / autonomous-vehicles-adaptive-cruise-control Goto Github PK