Pursuit and evasion games: semi-direct and cooperative control methods

摘要

Pursuit and evasion games have garnered much research attention since theclass of problems was first posed over a half century ago. With wide applicability toboth civilian and military problems, the study of pursuit and evasion games showedmuch early promise. Early work generally focused on analytical solutions to gamesinvolving a single pursuer and a single evader. These solutions generally assumed simple system dynamics to facilitate convergence to a solution. More recently, numericaltechniques have been utilized to solve more difficult problems. While many sophisticated numerical tools exist for standard optimization and optimal control problems,developing a more complete set of numerical tools for pursuit and evasion games isstill a developing topic of research.This thesis extends the current body of numeric solution tools in two ways.First, an existing approach that modifies sophisticated optimization tools to solvetwo player pursuer and evasion games is extended to incorporate a class of stateinequality constraints. Several classical problems are solved to illustrate the e?cacyof the new approach. Second, a new cooperation metric is introduced into the systemobjective function for multi-player pursuit and evasion games. This new cooperationmetric encourages multiple pursuers to surround and then proceed to capture anevader. Several examples are provided to demonstrate this new cooperation metric.