Skip to content

add Dominance-Informed Region Tree (DIRT)#1340

Open
aravindsiv wants to merge 2 commits intoompl:mainfrom
aravindsiv:main
Open

add Dominance-Informed Region Tree (DIRT)#1340
aravindsiv wants to merge 2 commits intoompl:mainfrom
aravindsiv:main

Conversation

@aravindsiv
Copy link
Copy Markdown

Adds Dominance-Informed Region Tree (DIRT) [1]. Similar to SST, DIRT is also an AO kinodynamic planner. Relative to SST:

  • DIRT has an informed implementation of node selection that biases selection towards high quality nodes close to the unexplored parts of the state space.
  • DIRT also evaluates multiple candidate controls as part of the node expansion. This Blossom expansion procedure ranks the resulting trajectories and processes them in order so as to prioritize expansions that bring the system closer to the goal according to a heuristic cost-to-go.

[1] Z. Littlefield and K. E. Bekris, "Efficient and Asymptotically Optimal Kinodynamic Motion Planning via Dominance-Informed Regions," 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Madrid, Spain, 2018, pp. 1-9, doi: 10.1109/IROS.2018.8593672.

@aravindsiv aravindsiv force-pushed the main branch 2 times, most recently from 6f5277d to 1dc510e Compare December 4, 2025 20:09
@ChamzasKonstantinos ChamzasKonstantinos self-assigned this Dec 7, 2025
@ChamzasKonstantinos
Copy link
Copy Markdown
Contributor

Hi @aravindsiv I can help review this pull request. Before I give you a detailed review, can you 1) Fix any CI issues left, I think it's your own test that is failing, and 2) add a demo script in C++ (ideally in python as well) that demonstrates how to use DIRT? Maybe a simple control problem that benchmarks DIRT against SST?

@aravindsiv
Copy link
Copy Markdown
Author

Thanks Constantinos, I hope my latest commit will fix the failing build. For the demo script, is the test sufficient, or can you point me to a similar script in the repo I can use as an example?

@ChamzasKonstantinos
Copy link
Copy Markdown
Contributor

Hi @aravindsiv since this is the geometric version of DIRT, I think it would make sense to add it here: https://github.com/ompl/ompl/blob/main/demos/OptimalPlanning.cpp , along with the geometric version of SST. Also just curious are you planning to also add the control version of DIRT? I think that one would be much more relevant, as the only AO control planner we have now is SST.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants