Augmented Reality and Deep Learning Guided Task Oriented Robot

Abstract

Robot navigation is one of the key features in robot system, which can be done in several ways. Different approaches show both strength and weakness, and the choice usually relies on the targeted scenario. The current work proposes a robot navigation system that works indoor for a certain specified task. In the system, augmented reality, SLAM (simultaneous localization and mapping) and deep learning are used to guide a robot. Augmented reality acts an interface bridging the physical and virtual world together. SLAM offers the robot the ability to sense the physical world. Deep learning enables the robot to recognize objects. As a result, a robot can follow the expected routine to get a task done without colliding.