PhotoI am a PhD candidate within the Algorithmics group at Delft University of Technology in the Netherlands. This website contains information on my research, publications, activities and other news you may be interested in.

News

Mar 2018
A paper describing our approximate algorithms for Constrained POMDPs will appear in the Journal of Artificial Intelligence Research (JAIR).

Jan 2018
Our work on improving exact algorithms for multi-objective and partially observable MDPs (with Diederik Roijers) will be presented at ICAPS.

Research

My research interests lie in the field of planning, scheduling, decision making under uncertainty, machine learning and artificial intelligence. Below you can find an overview of recent projects.

Wind power

Planning and scheduling in smart grids

My work focuses on planning and scheduling algorithms to control flexible loads in distribution grids, exploiting stochastic information on power generation and consumption. My research is part of the GCP project and the NWO URSES program. More info:

Smoover app

Smoover

We developed reinforcement learning algorithms to compute speed limits for highways. These algorithms have been used to support the development of Smoover, a new Intelligent Transportation System in the Netherlands which aims to reduce traffic congestion. More info:

Publications

Column Generation Algorithms for Constrained POMDPs

Erwin Walraven and Matthijs T. J. Spaan
Journal of Artificial Intelligence Research, 2018.

Bootstrapping LPs in Value Iteration for Multi-Objective and Partially Observable MDPs

Diederik M. Roijers, Erwin Walraven and Matthijs T. J. Spaan
Proceedings of the 28th Int. Conference on Automated Planning and Scheduling, pp. 218–226, 2018.

Accelerated Vector Pruning for Optimal POMDP Solvers

Erwin Walraven and Matthijs T. J. Spaan
Proceedings of the 31st AAAI Conference on Artificial Intelligence, pp. 3672–3678, 2017.

Bounding the Probability of Resource Constraint Violations in Multi-Agent MDPs

Frits de Nijs, Erwin Walraven, Mathijs M. de Weerdt and Matthijs T. J. Spaan
Proceedings of the 31st AAAI Conference on Artificial Intelligence, pp. 3562–3568, 2017.

Planning Under Uncertainty for Aggregated Electric Vehicle Charging with Renewable Energy Supply

Erwin Walraven and Matthijs T. J. Spaan
Proceedings of the 22nd European Conference on Artificial Intelligence, pp. 904–912, 2016.

Traffic flow optimization: A reinforcement learning approach

Erwin Walraven, Matthijs T. J. Spaan and Bram Bakker
Engineering Applications of Artificial Intelligence, vol. 52, pp. 203–212, 2016.

Planning under Uncertainty for Aggregated Electric Vehicle Charging using Markov Decision Processes

Erwin Walraven and Matthijs T. J. Spaan
International Workshop on Artificial Intelligence for Smart Grids and Smart Buildings, 2016.

Planning under Uncertainty with Weighted State Scenarios

Erwin Walraven and Matthijs T. J. Spaan
Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence, pp. 912–921, 2015.

A Scenario State Representation for Scheduling Deferrable Loads under Wind Uncertainty

Erwin Walraven and Matthijs T. J. Spaan
The 10th Annual Workshop on Multiagent Sequential Decision Making Under Uncertainty, 2015.

Traffic Flow Optimization using Reinforcement Learning (abstract)

Proceedings of the 26th Benelux Conference on Artificial Intelligence, pp. 211–212, 2014.

Traffic Flow Optimization using Reinforcement Learning

Master's thesis, Delft University of Technology, 2014.

Enhancing SAT Based Planning with Landmark Knowledge

Jan Elffers, Dyan Konijnenberg, Erwin Walraven and Matthijs T. J. Spaan
Proceedings of the 25th Benelux Conference on Artificial Intelligence, pp. 64–71, 2013.

Activities

Software

Our accelerated version of incremental pruning for POMDPs is available in SolvePOMDP, which is an open source software toolbox containing exact and approximate algorithms for POMDPs.

Contact

Van Mourik Broekmanweg 6
2628 XE Delft
The Netherlands
Phone: +31 (0)15 27 86206
Email:
Office: 3.E.280