Batch Informed Trees (BIT*) @ UTIAS / CMU (ICRA 2015)

This video shows an improvement to the popular RRT* planning algorithm that focusses the search area by eliminating some regions that cannot possibly improve the solution. It turns out we use an admissible ellipsoidal search region to do this, thereby greatly speeding up the search. Moreover, samples are drawn from the ellipse in batches and then ordered based on a heuristic. This has the effect that samples more likely to improve the path are processed first.
This video is associated with the following paper:
Gammell J D, Srinivasa S S, and Barfoot T D. “BIT*: Batch Informed Trees for Optimal Sampling-based Planning via Dynamic Programming on Implicit Random Geometric Graphs”. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA). Seattle, Washington, 26-30 May 2015.
asrl.utias.utor...