Home
example graphic

About Me

I recently completed my PhD at the Machine Learning Group at the University of Cambridge, supervised by Professor Carl Rasmussen and advised by Professor Zoubin Ghahramani. I am a member of King's College and was funded by the King's College Cambridge Studentship and University of Sydney Travelling Scholarship.

Prior to my PhD I was both an undergraduate and masters student at the Australian Centre for Field Robotics within the University of Sydney supervised by Dr Robert Fitch and advised by Dr Thierry Peynot. I worked on autonomous reconfiguration planning for modular robots and also a motion planning of a rover over unstructured terrain.

Résumé

Research Interests

I am interested in how learning probabilistic dynamics models enables data-efficient policy optimisation in reinforcement learning (RL). Currently I have been training a controller which makes use of a dynamics simulator and a filter that learns to balance the cart-pole system (starting with the pole in the down position) in the presence of significant observations noise. The method start with zero prior knowledge of the dynamics, yet learns to control the system within about 20 seconds worth of interaction. video[0.4MB]

I also have a general interest in RL and organised the Reinforcement Learning Reading Group within Cambridge's Machine Learning Group.

Contact

Publications
example graphic Concrete Problems for Autonomous Vehicle Safety: Advantages of Bayesian Deep Learning,
Rowan McAllister, Yarin Gal, Alex Kendall, Mark van der Wilk, Amar Shah, Roberto Cipolla, and Adrian Weller
International Joint Conference on Artificial Intelligence (IJCAI), 2017
example graphic Bayesian Learning for Data-Efficient Control,
Rowan McAllister
PhD thesis, University of Cambridge, 2017
example graphic Data-Efficient Reinforcement Learning in Continuous State-Action Gaussian-POMDPs,
Rowan McAllister and Carl Rasmussen
Neural Information Processing Systems (NIPS), 2017
example graphic Improving PILCO with Bayesian Neural Network Dynamics Models, [poster]
Yarin Gal, Rowan Mcallister, and Carl Rasmussen
Data-Efficient Machine Learning workshop, ICML, 2016
example graphic Data-Efficient Policy Search using PILCO and Directed-Exploration , [poster]
Rowan Mcallister, Mark van der Wilk, and Carl Rasmussen
Data-Efficient Machine Learning workshop, ICML, 2016
example graphic Learned Stochastic Mobility Prediction for Planning with Control Uncertainty on Unstructured Terrain,
Thierry Peynot, Angela Lui, Rowan McAllister, Robert Fitch, and Salah Sukkarieh
Journal of Field Robotics (JFR), 2014
youtube
example graphic Motion Planning and Stochastic Control with Experimental Validation on a Planetary Rover,
Rowan McAllister, Thierry Peynot, and Robert Fitch
International Conference on Intelligent Robots and Systems (IROS), 2012
example graphic Motion Planning and Stochastic Control with Experimental Validation on a Planetary Rover,
Rowan McAllister
Masters thesis, University of Sydney, 2012
example graphic Resilient Navigation through Online Probabilistic Modality Reconfiguration,
Thierry Peynot, Robert Fitch, Rowan McAllister, and Alen Alempijevic
International Conference on Intelligent Autonomous Systems, 2012
example graphic Autonomous Reconfiguration of a Multi-Modal Mobile Robot,
Thierry Peynot, Robert Fitch, Rowan McAllister, and Alen Alempijevic
ICRA workshop 2011
example graphic Hierarchical Planning for Self-Reconfiguring Robots Using Module Kinematics,
Robert Fitch, Rowan McAllister
International Symposium on Distributed Autonomous Robotics Systems (DARS 2010)
example graphic Autonomous Reconfiguration Planning in Modular Robots
Rowan McAllister
Honours thesis, University of Sydney, 2009
video[1.1MB] video[1.6MB] video[0.6MB]
Presentations