We introduce the Nav2 project's Smac Planner. It contains an optimized and templated A* implementation used for a broad range of planners, including: 2D-A*, Hybrid-A*, and State Lattice. These algorithms fill a broad need in the community for high-quality planning of circular and arbitrarily-shaped Differential, Omnidirectional, Ackermann, and Legged robots. The planners create kinematically feasible, continuous, and smooth paths with run-times around 10-200ms. Our State Lattice implementation allows for arbitrary control sets to support exotic motion-models, but we supply an auto-generator for typical models. This talk will go over key points of the planner, demonstrations, and how to enable it in an application.
Presented by: Steve Macenski (Samsung Research America)
