Improving the Generation of Rapidly Exploring Randomised Trees (RRTs) in Large Scale Virtual Environments Using Trails

摘要

Rapidly exploring randomised trees (RRTs) are a useful tool generating maps for use by agents to navigate. A disadvantage to using RRTs is the length of time required to generate the map. In large scale environments, or those with narrow corridors, the time needed to create the map can be prohibitive. This paper explores a new method for improving the generation of RRTs in large scale environments. We look at using trails as a new source of information for the agent's map building process. Trails are a set of observations of how other agents, human or AI, have navigated an environment. We evaluate RRT performance in two types of virtual environment, the first generated to cover a variety of scenarios an agent may face when building maps, the second is a set of 'real' virtual environments based in Second Life. By including trails we can improve the RRT generation step in most environments, allowing the RRT to be used to successfully plan routes using fewer points and reducing the length of the overall route.