Pre-contextualized Augmented Language Instructions for Autonomous Vision-and-Language Navigation

Advances in natural language has allowed for the automation of robotic tasks with minimal supervision, even in ambiguous situations. In the field of robotic navigation, human instruction can be utilized to allow a robotic agent to traverse unknown environments using instructional commands that mimic human-to-human interaction. This paper demonstrates the success of one such machine learning algorithms, delineating each machine learning architecture, and processes alternate fields where learning techniques can be applied for robotic automation.