Path Segmentation - Robotic Perception

This is an example of applied deep learning, specifically, semantic segmentation to estimate the drivable path for a robot in an unstructured environment.
The trained model used roughly 800 images (or less than one minute of labelled footage) to generalise well for use in this environment.

[@Vine-Robot]

Awesome stuff! May I ask how you created the pathing function for the robot given the segmented data coming from the video stream?