Evolution-based path planning and management for autonomous vehicles

This dissertation describes an approach to adaptive path planning based on the problem solving capabilities witnessed in nature—namely the influence of natural selection in uncovering solutions to the characteristics of the environment. The competition for survival forces organisms to either respond to changes or risk being evolved out of the population. We demonstrate the applicability of this process to the problem of finding paths for an autonomous vehicle through a number of different static and dynamic environments. In doing so, we develop a number of different ways in which these paths can be modeled for the purposes of evolution. Through analysis and experimentation, we develop and reinforce a set of principles and conditions which must hold for the search process to be successful. Having demonstrated the viability of evolution as a guide for path planning, we discuss implications for on-line, real-time planning for autonomous vehicles.