Hacker News Reader

This seems like another case of where the constant factors of these super-smart algorithms will end up nullifying any benefit you get, even producing worse performance.

View on HN

Like antics said, this isn't true for graph planning. Here are three concrete examples that I know of:

1. Anytime D* was used by the winner of the DARPA Urban Challenge

2. Field D* was used by Spirit and Opportunity on Mars

3. ARA* (and other A* variants) are a method of optimal motion planning in robotic manipulation

This is actually completely false. Most of these algorithms were specifically designed to make a specific intractable subcase of some algorithm tractable, not only in theory, but in practice. Most of them are completely practical, and together they constitute a really large portion of what we know about motion planning.

Reply on news.ycombinator.com