Dr Zeyn Saigol
Edinburgh, EH14 4AS, UK
My academic interests are in AI, and particularly in how to model the world in a probabilistic framework, machine learning, planning under uncertainty and robotics.
I work with David Lane on various projects in the OSL, including the European FP7 project PANDORA.
Previously I worked on the GeRT project at Birmingham, which is creating smart software to make humanoid robots capable of handling novel objects in an office or household setting. In collaboration with Richard Dearden I developed ways of learning generic actions from example task scripts, which is part of Work Package 1 of the project.
My PhD was supervised by Jeremy Wyatt and Richard Dearden, and was on planning under uncertainty in the domain of autonomous underwater vehicles (AUVs). The aim was for the submersible robot to make decisions to maximise the scientific reward obtained from a mission – for example, it should weigh up the probability of finding another site of interest against the risk of running out of battery before it is able to return to the surface. There are some movies of my novel algorithms in action available (DIVX encoded). This work was in collaboration with Bramley Murton from the National Oceanography Centre, Southampton.
Chris Burbridge, Zeyn A Saigol, Florian Schmidt, Christoph Borst, and Richard W Dearden
Learning Operators for Manipulation Planning
In Proceedings of IROS 2012. October 2012.
Francesco Maurelli, Zeyn A Saigol, Carlos C Insaurralde, Yvan R Petillot, and David M Lane
Marine World Representation And Acoustic Communication: Challenges For Multi-Robot Collaboration
In Proceedings of 2012 IEEE/OES Autonomous Underwater Vehicles (AUV2012). September 2012.
Francesco Maurelli, Zeyn A Saigol, Joel Cartwright, David M Lane, Alex Bourque, and Bao Nguyen
TDMA-Based Exchange Policies For Multi-Robot Communication Of World Information
In Proceedings of IFAC MCMC 2012 - Manoeuvring and Control of Marine Craft. September 2012.
Zeyn A Saigol, Richard W Dearden, Jeremy L Wyatt, and Bramley J Murton
Belief Change Maximisation for Hydrothermal Vent Hunting Using Occupancy Grids
In Proceedings of TAROS 2010. September 2010.
Zeyn Saigol, Frédéric Py, Kanna Rajan, Conor McGann, Jeremy Wyatt, and Richard Dearden
Randomized Testing for Robotic Plan Execution for Autonomous Systems
In Proceedings of 2010 IEEE/OES Autonomous Underwater Vehicles (AUV2010). September 2010.
Zeyn A Saigol, Richard W Dearden, Jeremy L Wyatt, and Bramley J Murton
Information-Lookahead Planning for AUV Mapping
In Proceedings of IJCAI 2009. July 2009.
Also a long version of the paper, published as Tech Report CSR-09-01, which has colour versions of the figures and some extra derivations.
Richard W Dearden, Zeyn A Saigol, Jeremy L Wyatt, and Bramley J Murton
Planning for AUVs: Dealing with a Continuous Partially-Observable Environment
In Workshop on Planning and Plan Execution for Real-World Systems, ICAPS 2007. September 2007.
Tim L Phillips, Zeyn A Saigol, and Simon Hanna
Computer Simulation of Rotator Phases in Liquid-Crystalline Polymers
In Molecular Crystals and Liquid Crystals, Volume 303, Issue 1, pages 9 - 14. September 1997.
Zeyn A Saigol
Automated Planning for Hydrothermal Vent Prospecting Using AUVs
(Also a screen-friendly version with hyperlinks and less spacing)
PhD Thesis, IRLab, University of Birmingham. April 2011.
Reports prepared for my thesis group (which includes Xin Yao and Hamid Dehghani in addition to my supervisors, Richard Dearden and Jeremy Wyatt).
Report 8 (April 2010)
Report 7 (October 2009)
Report 6 (April 2009)
Report 5 (October 2008)
Report 4 (April 2008)
Report 3 (Thesis Proposal) (September 2007)
Report 2 (April 2007)
Report 1 (November 2006)
Justin simulator and GeRT action learning presentation
given to the
IRLab (Mar 2011)
TAROS 2010 slides (PDF) from presenting Belief Change Maximisation for Hydrothermal Vent Hunting Using Occupancy Grids paper (August 2010)
Submodularity overview (PDF), short IRLab presentation on Andreas Krause and Carlos Guestrin's work on leveraging submodularity for choosing sensor locations, based on their slides (July 2010)
Using entropy to drive search in occupancy grids, update on my PhD work given to the Intelligent Robotics Lab (February 2010)
IJCAI 2009 slides from presenting Information-Lookahead Planning for AUV Mapping paper (July 2009)
Overview of Pearl's causation work (see Pearl's homepage), given to the Intelligent Robotics Lab (March 2009)
Postgraduate Seminar Series talk covering my PhD problem, early results, and my internship at MBARI (January 2009)
Approximate approaches to solving grid-world POMDPs for AUVs, overview of my PhD progress given to the Intelligent Robotics Lab (December 2008)
Automated testing for AUV planning software, presentation of my internship project at MBARI (August 2008)
Presentation of Kaelbling, Littman and Cassandra's POMDP / Witness paper, given to the Intelligent Robotics Lab (March 2008)
Why has AI not (yet) succeeded? – part of the cake talk series organised by the PhD students in my department (January 2008)
Robot mapping overview (PDF) given to my research group (the Intelligent Robotics Lab) (October 2007)
ICAPS 2007 Workshop on Planning and Plan Execution for Real-World Systems talk, supporting the paper (September 2007)
Intelligent Robotics Lab presentation given to my research group in the school (July 2007)
Short presentation given to other first year PhDs in the school, describing the aims of my PhD (January 2007)
A Java GraphPlan implementation, which takes problems defined in PDDL as input.
Software for fitting an Integrate-and-Fire model to neural data, using a linear-regression-based technique. The code is written in MATLAB, and was developed for my Edinburgh MSc project.
I thoroughly recommend Lyx, an open-source editor that liberates you from having to write raw LaTeX.
For 10 weeks over summer 2008, I was an intern at MBARI in California. Working under Kanna Rajan, I developed enhancements to the constraint-based planning software that runs on MBARI's AUVs, and wrote a Monte-Carlo test harness to help improve the robustness of the system to different ocean environments. My internship report is available, as well as a local cache of roughly what my MBARI webpage looked like.
I completed an MSc at Edinburgh in September 2006, where my dissertation was on numerical optimisation of neural models, supervised by Chris Williams and co-supervised by David Sterratt.
University of Edinburgh, MSc in Informatics (with distinction)
Specialism: Learning from Data
Before that I was a software developer, mostly writing web-based Java applications with RDBMS back-ends. I worked for the following companies (in reverse chronological order):
Pentacom Communications (now defunct)
Tertio (now merged into Evolving Systems)
SSA Softwright (now merged into Syniverse Technologies)
My first round in higher education consisted of:
Imperial College London, MSc in Computing (with distinction)
Dissertation: A C to C++ Converter, describing a refactoring tool for converting structs and functions in C programs into classes and member functions in C++ code.
Supervisor: Sophia Drossopoulou
University of Bristol, BSc in Physics (I)
Final year project: Simulation of a liquid crystal polymer