An Ego-Centric Interface for Human Intervention in Semi-Autonomous Navigation through Unknown Environments

Abstract

Semi-autonomous navigation through unknown environments is an important area of research for robotics, with many potential applications in fields such as search and rescue, environmental monitoring, and exploration. However, existing methods rely on the robot to already have a semi-accurate map of the environment which cannot be assumed to be true in an unfamiliar environment. To address this, a new egocentric interface has been presented in this research paper, which provides a Robot Operating System (ROS)-based software utilizing the robot’s camera view. With this interface, a human operator can select a point visible in the robot’s camera view and provide it as an intermediate goal in the robot’s local environment. Using the robotic system’s inherent path-planning and navigation tools, a collision-free path is derived for the robot to follow. While the robot moves along the given path, a human operator may continue to provide intermediate goals and adjust its path based on recent observations. This process is simulated through the Gazebo simulator.