self-driving-car-predict-steering

Predicting steering angle for self driving car using deep learning

pre-requisites:

1. install ros
2. install numpy
3. install matplotlib
4. install pandas
5. put the bag file in the dataset folder with name "dataset.bag"

visualise the camera input

run roslaunch dataprocess visualise_camera_images.launch
this should launch rviz displaying the camera data.

getting the data out of bag file

for generating images from cameras and csv files
run python bag_extract_data.py
this should extract all the images into corresponding folders left, right, center and the folder yaml_files will contain the required csv files, the final_interpolated.csv will contain all image file names with corresponding data (steering angle, acceleration etc.) values.

for only generating image files
run python bag_extract_image_data.py
this will extract all the images into corresponding folders left, right, center and genrate a text file image_steering_angle.txt with values

training

for training on center camera images
run python train_center.py
change the epochs and batch size accordingly

for training using all the three camera images
run python train_merged.py
change the epochs and batch size accordingly

testing

run inference.py this should load the weights and predict steeering angle on the test data

Notes:

1. The .yaml, .png, .bag files are ignored in git