0.3.0

New features:

• Support for joystick control;

• Colour mask for signal detection;

• Bug fixes;

Features that need attention:

• Template matching is slow;

0.2.1

New release used to correct bugs:

• Launch files not working due to leftover parameters;

And to add:

• Log files for signal detection;

0.2.0

This release added the following features:

• Full integration of the code with Kinect V1 as the lane camera;
• Functional signal recognition, except on cnn2av;
• First developments on inverse perspective mapping;
• Creation of an .yaml file after dataset gathering and neural network training, in order to retrieve information about said database or neural network model;

v0.1.1

This release includes the following functions:

• Fully simulated virtual environment using Gazebo;
• Acckermann car simulated on that same environment;
• MCU code to control an ESC and an acckermann steering direction, using ROS subscribers;
• Convert webcam image to a ROS topic;
• Record the robot view as .jpg images;
• Corresponds those .jgp images with the velocity and angular direction commands the robot was receiving;
• With the previous data, trains a ML model, using one of two different CNNs;
• Using the ML model, drives autonomously;
• Recognizes signals and drives/stops accordingly - Only on virtual environment;
• .launch files for every functionality we have, to reduce the number of commands needed;