Path-aware frontier ranking shortens LiDAR drone exploration flights by up to 14%

Researchers led by Duc Manh Do have described in an arXiv paper a light method that helps LiDAR-equipped drones choose where to explore next. Ranking unexplored frontiers by straight-line distance can misjudge the real collision-free flight cost in cluttered spaces, so the method keeps a few promising candidates, plans time-limited RRT* paths to them to estimate the true cost, and reuses the chosen path for navigation.

In three warehouse-like environments, it produced shorter exploration flights than conventional frontier methods, cutting the distance travelled by up to 14.1% and exploration time by up to 9.5% in two of them, while keeping map detail similar to an OctoMap-based baseline.