Estimation of transitability based on monocular images

The project aims to implement a system that allows the robot called "Chimuelo" the necessary autonomy to analyze and execute a displacement from a starting point called A to a final one called B in the most efficient and optimal way possible, through of artificial vision and deep learning. Likewise, a program will be developed that from the segmentation will provide the robot with control slogans (controller).

It should be noted that this system has as an additional objective that from monocular images the transitability of an environment can be estimated in order to calculate control slogans for a terrestrial robot.

This project will be compiled in a ROS package and in a graphics processing unit called Jet2 TX2 GPU that ensures a performance superior or equal to 10 frames per second.