Class EIRMstar
Defined in File EIRMstar.h
Inheritance Relationships
Base Type
public ompl::geometric::EITstar(Class EITstar)
Class Documentation
-
class EIRMstar : public ompl::geometric::EITstar
Effort Informed Roadmaps (EIRM*)
EIRM* (Effort Informed Roadmaps) is an almost-surely asymptotically optimal multiquery path planner based on EIT*. It aims to find an initial solution quickly by resuing previously invested computational effort (i.e., validated edges) and asymptotically converges to the globally optimal solution.
It uses an asymmetric bidirectional search to identify known valid edges in a lazy roadmap and then reuses these to quickly solve individual planning queries by reducing computational effort. As EIT*, EIRM* then adds batches of samples to its approximation when improving the solution. To avoid issues that come with the growing graph size from refining the approximation to find the globally optimal solution, EIRM* resets its approximation to the initial batch of samples at each new query in a multiquery problem.
V. N. Hartmann, M. P. Strub, M. Toussaint, J. D. Gammell. “Effort informed roadmaps (EIRM*):
Efficient asymptotically optimal multiquery planning by actively reusing validation effort” Submitted to Proceedings of the International Symposium on Robotics Research (ISRR) 2022
- Associated publications:
arXiv: arXiv:2205.08480 Video 1: ISRR trailer