Understanding Linear Quadratic Drone Games through Simulation: Linear Quadratic Games as a Baseline for Evaluating Multi- Agent Reinforcement Learning Algorithms and Simulation as a Tool for Understanding and Innovation

Sammanfattning

A relevant method of generating control policies for autonomous drones is throughmulti-agent reinforcement learning (MARL) algorithms. Novel MARL algorithmsdo not always have theoretical guarantees of convergence which motivates the needfor robust and reliable baselines to benchmark these algorithms against. This thesisinvestigates the efficacy of derived Nash equilibrium (NE) solutions to the familyof linear quadratic (LQ) games as one such possible baseline. The first part ofthe thesis derives the Nash equilibrium solutions for LQ games, which serve as thebaseline policies. Based on these results, two experimental scenarios are designedto benchmark MARL algorithms against the baseline. The experimental resultsindicate that the MARL policies perform on par with the baseline in the two-playerscenario, while outperforming the baseline in the cooperative five-player scenario.The second part of the thesis explores to what extent computer simulation can beused as an effective knowledge sharing method at an engineering company. Anexploratory pilot study was conducted comparing two learning sessions, one sessionemploying an online simulation tool developed for this purpose and the other being atraditional lecture. Responses collected after each learning session indicate that thevisual and interactive elements of the simulation tool were conducive to generatingengagement and curiosity among participants. Furthermore, providing necessarycontext and examples of applicability were deemed important aspects when sharinginformation about a novel topic among engineers.