Alvaro Cabrejas Egea

I am a PhD Graduate from the MathSys CDT. I passed my viva on January 2021. I am currently based in the Alan Turing Institute in the British Library (London). Normally, I can be found in the first floor.

My PhD project is centred around transportation systems modelling, prediction and control, using a variety of tools to create novel applications of machine learning and time series analysis. Currently I make myself busy trying to put little brains inside traffic lights and teaching them to optimise vehicular and pedestrian flow using Reinforcement Learning.

Recent and Current Projects

Industrial Internship with Vivacity Labs. Looking into Reward Function for Reinforcement Learning agents for Urban Traffic Control.
Internship at the Alan Turing Institute. Optimising flow within mobility systems with AI (Reinforcement Learning), in a collaboration with the Toyota Mobility Foundation. Working with Sir Alan Wilson (UCL, ATI), Neil Walton (Manchester) and Damon Wischik (Cambridge).
PhD on Traffic Management and Prediction with Dr Colm Connaughton. The project is a continuation of the Research Group Project. Initially we are focusing on the limitations of forecasting based on profiling time series. Next steps will involve Shockwave Absorption, Data Assimilation and Fluid Dynamics models.
Research Group Project with Ayman Boustati, Peter De Ford, Laura Guzman and Guillem Mosquera on Intelligent Mobility Applications for the UK Strategic Road Network. This project is in collaboration with the THALES Group and is under the supervision of Dr Colm Connaughton.
MSc Project on Simulation Optimisation with Simulated Annealing. In collaboration with Lanner and under the supervision of Dr Robin Ball and Dr Juergen Branke.
Research project on sociophysics titled The Rise of Collaboration: Convergence in Social Networks with Directed Punishment with Samuel Johnson.
Smart Sustainable Cities project with Siemens Traffic in the Alan Turing Institute, London. It involved deploying independent Q-Learning Agents in each intersection of the Coventry city centre road map.

Publications

Assessment of Multi-Objective Reward functions for Reinforcement Learning Traffic Signal Control under Real-World limitations, A. Cabrejas Egea, C. Connaughton. Under review IEEE ITSC2021.
Pre-print: https://arxiv.org/abs/2010.08819
Reinforcement Learning for Traffic Signal Control: Comparison with Commercial Systems, A. Cabrejas Egea, R. Zhang, N. Walton. Accepted to CIT2021. Pre-print: https://arxiv.org/abs/2104.10455
Assessment of Vehicular Reward functions for Reinforcement Learning Traffic Signal Control under Real-World limitations, A. Cabrejas Egea, S. Howell, M. Knutis, C. Connaughton. IEEE International Conference on Systems, Man, and Cybernetics 2020. https://ieeexplore.ieee.org/document/9283498 (Open access version in ArXiv).
Wavelet Augmented Regression Profiling (WARP): improved long-term estimation of travel time series with recurrent congestion. A. Cabrejas Egea, C. Connaughton. In 23rd IEEE International Conference on Intelligent Transportation Systems (ITSC2020), Pre-print: https://arxiv.org/abs/2006.13072.
Design choices for productive, secure, data-intensive research at scale in the cloud. A. Cabrejas Egea et al. (no specific order), Aug 2019 https://arxiv.org/abs/1908.08737
Mitigation of Operational Risks in Banking with Machine Learning. Alan Turing Institute, Accenture, Global Bank. Jan 2019.
Estimating Baseline Travel Times for the UK Strategic Road Network, A. Cabrejas Egea, P. de Ford Gonzalez, C. Connaughton. In 2018 21st International Conference on Intelligent Transportation Systems (ITSC), pp. 531-536. IEEE, 2018. https://ieeexplore.ieee.org/document/8569924
New Applications of Data Science to Intelligent Transportation Systems, A. Cabrejas Egea. PhD Thesis. Jan 2021. Undergoing minor corrections.
Discount Factor and Temporal Timescales in Reinforcement Learning for Urban Traffic Control, A. Cabrejas Egea, C. Connaughton [Writing up]
Minimizing Pollution and Emissions in Urban Traffic Control. Siemens and Alan Turing Institute. [Pending publication]

Previous Projects

Robotics projects designing and building electromechanical systems and implementing their Control Systems.
Different projects in Aerospace Engineering mostly centered on the Primary Structure of Airbus A350 & A380.

Undergraduate Teaching

Graduate Teaching

Education and experience

2007-2012 BSc Aerospace Engineering: Specialisation in Aircrafts - Universidad Politecnica de Madrid
2012-2015 Professional Experience in Aerospace Sector (see LinkedIn)
2015-2016 MSc Mathematics for Real-World Systems - University of Warwick
2017-2018 PhD Mathematics for Real World Systems - University of Warwick
2018-2019 Enrichment Year PhD Student - Alan Turing Institute (London)
2019-2021 PhD MathSys at University of Warwick & Science Liaison at Alan Turing Institute

Hobbies

In my spare time I enjoy swimming, reading, skimboarding, bodyboarding, playing basketball, grand strategy games and travelling.

Research Interests

Machine Learning and Reinforcement Learning
Optimisation, Control and Operations Research
Swarm Dynamics and Collective Behaviour
Economic and Agents-Based Games in Networks

Contact

a.cabrejas-egea [AT] warwick.ac.uk

acabrejasegea [AT] turing.ac.uk