3D Perception for Autonomous Robot Exploration

• Published in: Advances in Visual Computing pp. 888 - 900
• Main Authors: Xu, Jiejun, Kim, Kyungnam, Zhang, Lei, Khosla, Deepak
• Format: Book Chapter
• Language: English
• Published: Cham Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Autonomous Robot Exploration; Input Point Cloud; Obstacle Avoidance Module; Proposed Algorithm; Vanishing Points
• ISBN: 9783319278568; 3319278568
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-27857-5_79

We propose an online 3D sensor-based algorithm for autonomous robot exploration in an indoor setting. Our algorithm consists of two modules, a proactive open space detection module, and a reactive obstacle avoidance module. The former, which is the primary contribution of the paper, is responsible for guiding the robot towards meaningful open spaces based on high level navigation goals. This generally translates to identifying open doors or corridor vanishing points in a typical indoor setting. The latter is a necessary component that enables safe autonomous exploration by preventing the robot from colliding with objects along the moving path. Assuming a 3D range sensor is mounted on the robot, it continues to scan and acquire signal from its surroundings as it explores in an unknown environment. From each 3D scan, the two modules function cooperatively to identify any open spaces and obstacles within the generated point cloud using robust geometric estimation methods. Combination of the two modules provides the basic capability of a autonomous robot to explore an unknown environment freely. Experimental results with the proposed algorithm on both real world and simulated data are promising.