1. PDRRTs: Integrating Graph-based and Cell-based Planning
    Ananth Ranganathan and Sven Koenig, IROS 2004

    When a robot is given an opportunity to perform motion planning and path-finding in the same environment repeatedly, it should take advantage of this by learning about the environment and getting better at these problems with time. In spaces of small dimensionality or when a map of the environment is given, standard motion planning algorithms such as RRTs can provide good solutions. However, in high-dimensional spaces which are unknown apriori, these do not work. This paper provides a novel algorithm for learning unknown environments by solving multiple path-finding problems in them.