Leveraging Factored State Representations for Enhanced Efficiency in Reinforcement Learning

Reinforcement learning techniques have demonstrated great promise in tackling sequential decision-making problems. However, the inherent complexity of real-world scenarios presents significant challenges for its application. This thesis takes a fresh approach that explores the untapped potential of factored state representations as a means to...

Abstraction-Guided Modular Reinforcement Learning

Reinforcement learning (RL) models the learning process of humans, but as exciting advances are made that use increasingly deep neural networks, some of the fundamental strengths of human learning are still underutilized by RL agents. One of the most exciting properties of RL is that it appears to be incredibly flexible, requiring no model or...

Risk Aversion and Guided Exploration in Safety-Constrained Reinforcement Learning

In traditional reinforcement learning (RL) problems, agents can explore environments to learn optimal policies through trials and errors that are sometimes unsafe. However, unsafe interactions with environments are unacceptable in many safety-critical problems, for instance in robot navigation tasks. Even though RL agents can be trained in...

Safe Online and Offline Reinforcement Learning

Reinforcement Learning (RL) agents can solve general problems based on little to no knowledge of the underlying environment. These agents learn through experience, using a trial-and-error strategy that can lead to effective innovations, but this randomized process might cause undesirable events. Therefore, to enable the adoption of RL in our...

Solving Large-Scale Dynamic Collaborative Vehicle Routing Problems: An Auction-Based Multi-Agent Approach

The freight transportation sector is one of the major contributors to air pollution. An important way to reduce emissions consists of collective route planning. Although unloaded trips and inefficient routes could not always be prevented by individual carriers, more efficient operations could often be obtained if multiple carriers collaborate by...

Collective Decision Making through Self-regulation: Mechanisms and Algorithms for Self-regulation in Decision-Theoretic Planning

This thesis explores the potential of self-regulation in collective decision making to align interests and optimise joint performance. Demonstrated in the domain of road maintenance planning, this research contributes novel incentive mechanisms and algorithmic techniques to incite self-regulation and coordinate agent interactions, paired with a...

DC Distribution Systems: Modeling, Stability, Control & Protection

Historically speaking, alternating current (ac) has been the standard for commercial electrical energy distribution. This is mainly because, in ac systems, electrical energy was easily transformed to diffierent voltages levels, increasing the efficiency of transmitting power over long distances. However, technological advances in, for example,...

Planning under Uncertainty in Constrained and Partially Observable Environments

Developing intelligent decision making systems in the real world requires planning algorithms which are able to deal with sources of uncertainty and constraints. An example can be found in smart distribution grids, in which planning can be used to decide when electric vehicles charge their batteries, such that the capacity limits of lines are...

Resource-constrained Multi-agent Markov Decision Processes

Intelligent autonomous agents, designed to automate and simplify many aspects of our society, will increasingly be required to also interact with other agents autonomously. Where agents interact, they are likely to encounter resource constraints. For example, agents managing household appliances to optimize electricity usage might need to share...