GePA*SE: Generalized Edge-Based Parallel A* for Slow Evaluations

TL;DR

GePA*SE is a new parallel search algorithm that efficiently handles domains with actions requiring varying computational efforts, extending previous methods to more diverse robotic planning scenarios.

Contribution

It introduces a generalized parallel A* algorithm that combines state expansion and edge evaluation parallelization for heterogeneous action costs.

Findings

Successfully extends parallel A* to heterogeneous action evaluation times.

Achieves improved planning speed in robotics domains with diverse action costs.

Open-source implementation available for benchmarking.

Abstract

Parallel search algorithms have been shown to improve planning speed by harnessing the multithreading capability of modern processors. One such algorithm PA*SE achieves this by parallelizing state expansions, whereas another algorithm ePA*SE achieves this by effectively parallelizing edge evaluations. ePA*SE targets domains in which the action space comprises actions with expensive but similar evaluation times. However, in a number of robotics domains, the action space is heterogenous in the computational effort required to evaluate the cost of an action and its outcome. Motivated by this, we introduce GePA*SE: Generalized Edge-based Parallel A* for Slow Evaluations, which generalizes the key ideas of PA*SE and ePA*SE i.e. parallelization of state expansions and edge evaluations respectively. This extends its applicability to domains that have actions requiring varying computational effort to evaluate them. The open-source code for GePA*SE along with the baselines is available here: https://github.com/shohinm/parallel_search