Goal-driven Mobile Robot Navigation in Unknown Indoor Environments

Date
2013-01-18
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
A goal-driven mobile robot navigation system is proposed for unknown indoor environments. The proposed system can be subdivided into four main modules, namely, localization, mapping, motion control, and goal detection. In the proposed system, a mobile robot uses an odometry system and a Kinect sensor as its input devices. An optimal particle filter models the posterior over the robot trajectory while minimizing the variance of the importance weights of the particles. Occupancy grid maps are employed to represent the environment as they do not make any assumptions on distinguishable landmarks. A nearness-diagram reactive navigation technique generates motion commands based on the robot position and navigation goal location. A trajectory parameter space is used as an abstraction layer of the robot shape and kinematic constraints for the nearness-diagram method. A goal-driven situation assessment framework based on fuzzy cognitive maps is developed to verify navigation goals using sensory information and expert knowledge.
Description
Keywords
Engineering--Electronics and Electrical, Robotics
Citation
Marasinghe Arachchige, N. G. (2013). Goal-driven Mobile Robot Navigation in Unknown Indoor Environments (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/26311