Semantic Labeling of Objects in a Simulated Environment

摘要

Situational awareness and trust are essential to promote seamless interactions between a robot and human. Perception of the world by each of these is through different lenses. Robots need a better understanding of the world through classification of the scene and semantic understanding of objects in the local environment. Perception algorithms provide a way for robots to gain knowledge of the environment. Training and testing of these algorithms take significant effort on the part of researchers. With the development of a simulation tool, researchers are now able to systematically validate their perception algorithms in a simulated environment before conducting research in the field.