Date

2016

Document Type

Dissertation

Degree

Doctor of Philosophy

Department

Computer Science

First Adviser

Spletzer, John

Other advisers/committee members

Chen, Brian; Huang, Xiaolei; Derenick, Jason

Abstract

Lifelong navigation of mobile robots is to ability to reliably operate over extended periods of time in dynamically changing environments. Historically, computational capacity and sensor capability have been the constraining factors to the richness of the internal representation of the environment that a mobile robot could use for navigation tasks. With affordable contemporary sensing technology available that provides rich 3D information of the environment and increased computational power, we can increasingly make use of more semantic environmental information in navigation related tasks.A navigation system has many subsystems that must operate in real time competing for computation resources in such as the perception, localization, and path planning systems. The main thesis proposed in this work is that we can utilize 3D information from the environment in our systems to increase navigational robustness without making trade-offs in any of the real time subsystems. To support these claims, this dissertation presents robust, real world 3D perception based navigation systems in the domains of indoor doorway detection and traversal, sidewalk-level outdoor navigation in urban environments, and global localization in large scale indoor warehouse environments.The discussion of these systems includes methods of 3D point cloud based object detection to find respective objects of semantic interest for the given navigation tasks as well as the use of 3D information in the navigational systems for purposes such as localization and dynamic obstacle avoidance. Experimental results for each of these applications demonstrate the effectiveness of the techniques for robust long term autonomous operation.

Share

COinS