feng3245 / carnd-carla

Udacity's Self-Driving Car Capstone Project: Programming a Real Self-Driving Car

SDC Fun

This is the SDC Fun team repo for the capstone project of Udacity's Self-Driving Car Engineer Nanodegree. The team members are:

• James Barfield ([email protected])
• Nick Condo([email protected])
• Jim Duan ([email protected])
• Nimish Sanghi ([email protected])
• Colin Shaw ([email protected])

Building and running the project

The original project repo can be found here, which has a lot of information about environment, simulator, etc.

This project requires the GPU version of Tensorflow along with the CUDA and cuDNN libraries, python 2.7, as well as the following libraries defined more specifically in requirements.txt:

• Flask
• attrdict
• eventlet
• python-socketio
• numpy
• Pillow
• scipy
• keras
• h5py
• requests

These modules can be installed using: pip install -r requirements.txt from the root of the project.

You will need to download our model in order to detect traffic lights. You can download the model by navigating to self-driving-car-capstone/data_science/models and running python get_model.py. You may have to manually install requests since it is not included in the upstream requirements.txt and is needed for downloading the model. Alternatively, you can download the model manually here.

Next go into the /ros directory and run catkin_make to build the ROS project. Be sure to source the project by running . devel/setup.sh. At this point, if your environment is set up properly, you should be able to launch ROS with roslaunch launch/styx.launch.

Further information about our implementation can be found here.