Skip to end of metadata
Go to start of metadata

Timothy Morris

Contact Details

Timothy Morris | PhD Candidate
School of Electrical Engineering and Computer Science
Queensland University of Technology
Gardens Point, S Block 1107 | 2 George Street, Brisbane, QLD 4000
email: timothy.morris@qut.edu.au

PhD Information

Title: Lifelong robotic navigation using visual perception
Supervisor:  Dr. Ben Upcroft
Associate Supervisors: Prof. Peter Corke, Dr. Feras Dayoub

Brief bio

Timothy Morris received his Bachelor of Mathematics in 2010 and Bachelor of Information Technology (Computing Science, Hons I) in 2011 from Queensland University of Technology. During his time as an undergraduate he worked on image processing problems including a vision sensing hexapod and a shape based classifier for visual traffic monitoring.

Current Research

Spatio-Temporal Framework for Localisation and Mapping

Existing Simultaneous Localisation and Mapping (SLAM) techniques operate well over short periods of time, there is a need to consider temporal changes such as lighting and object displacement to achieve long-term accuracy. The goal of this research is to develop a framework that can combine local spatial representations of an environment separated by different times / conditions into a global map while maintaining global consistency. Focus will be given to a feature agnostic approach that can recover from erroneous additions to the map by exploring a multiple hypotheses representation through active navigation. This framework is a partial contribution to a group effort, attempting to solve the problem of lifelong robotic navigation using visual perception: Persistent Robotic Navigation being challenged by the CyPhy Lab at QUT.

Massively Sparse 3D Representations

Most commonly, 3D representations of environments are used for human visual interpretation, sparse representations require surface like properties for us to identify with. Using a massively sparse set of points taken with stereo vision, the goal is to identify the minimum set that can be used to describe a position in space of varying appearance. Such spatial information can aid in better localisation within maps created of dynamic environments.

Experimentation

An Adept Guiabot (named "Casper") is used to conduct experiments towards lifelong navigation using visual perception. This robot has been modified to support a stereo rig of Point Grey 2/3" CCD Grasshopper cameras, Point Grey Ladybug omnidirectional camera and a Kinnect sensor. Processing is handled on-board by two i7 Quad Core Mac Mini's.

The robot operated for 2 months during business hours on level 11 S-Block and amongst the crowds that attended Robotronica 2013 (approximately 24, 000 people in attendance). We are currently conducting long-term experiments on level 3 P-Block and inside the QUT Bookshop. These experiments are designed to test our multiple hypotheses approach in a highly dynamic environment where efficient paths can change over time. The paths chosen by Casper under different modes of operation will be compared over a long period, along with assessments of crowd density made by the human supervisor. This information will be used to measure navigation performance of a robot that adapts to its environment verses one that does not. See The Guiabot page for further details.

Publications

Google Scholar

F. Maire, T. Morris, A. Rakotonirainy "Segmentation of scenes of mobile objects and demonstrable backgrounds", In proceedings of the 6th International Symposium on Autonomous Minirobots for Research and Edutainment (AMiRE), Bielefeld, May 2011.

L. Murphy, T. Morris, U. Fabrizi, et al. "Experimental comparison of odometry approaches", In proceedings of the International Symposium on Experimental Robotics (ISER), Quebec 2012.

L. Murphy, T. Morris, M. Milford, et al. "A Framework for Mix & Match Vision-Only Navigation", Queensland University of Technology - CyPhy lab, Technical Report, 2012.

F. Dayoub, T. Morris, B. Upcroft, P. Corke, “Vision-Only Autonomous Navigation Using Topometric Maps”, International Conference on Intelligent Robots and Systems (IROS) 2013.

F. Dayoub, T. Morris, B. Upcroft, P. Corke, “One robot, eight hours, and twenty four thousand people”, Australian Conference on Robotics and Automation (ACRA) 2013.

T. Morris, F. Dayoub, P. Corke, G. Wyeth, B. Upcroft, “Multiple map hypotheses for planning and navigation in a non-stationary environment”, International Conference on Robotics and Automation (ICRA) 2014 (In Review).