arXiv preprint ·
Does RRT* Work on Uneven Terrain Meshes?
- Published
- Source
- arXiv preprint, posted before peer review
- Original
- Read it at arXiv
- Topics
Abstract
The optimal rapidly exploring random tree (RRT*) is proven asymptotically optimal on open subsets of Euclidean space with straight-line local planning. We ask whether it works on a triangle-mesh terrain, a two-dimensional surface in R^3. First, the original RRT* cannot run on a mesh: the surface has zero volume, so uniform sampling never hits it, no clearance ball fits in it, and chords across a fold leave it. Second, the surface reformulation we call Mesh RRT*, with its naive edge, the chord costed by its length, solved every run on heightfields with slopes up to 75 degrees but minimises the wrong objective: its cost, the straight-line length of a polyline whose vertices lie on the surface while its segments may leave it by a tolerance tau, is the length of no route on the surface and lies strictly below the geodesic length on any curved mesh. In numeric simulation and after 30,000 iterations the reported cost is on average up to 2.7% below the exact geodesic, a cost no path attains, and the returned route on average 0.1-0.5% longer than it. Rewiring keeps the biased edges, the bias decays only linearly with the step on polyhedral surfaces, and shrinking tau trades bias for connectivity. Third, replacing each chord by a path on the surface above it, built from projected points joined across the mesh edges they straddle and costed by its own length, makes every edge cost the exact length of a feasible path; with no other change the exact length of the returned routes is on average within 0.07-0.16% of the geodesic at every roughness tested, two to 3.4 times closer than with chords on the rougher terrains. The planner is in MeshNav3D 0.2.0 (release pending).
The authors’ abstract, as arXiv lists it. arXiv’s listings, abstracts included, are free for anyone to reuse.
Read it on ReadSearch
Each opens after you create an account, and brings you back to this paper.
More in Robotics
Softmax Attention with Any Temperature in Fixed Dimension: Matching the Best Known Maximum-Inner-Product Bounds
Constant-Round Massively Parallel Algorithms for Approximate MSTs in Metric Spaces
Resolving Tusnády's Problem
Stacking Stars: Minimum-Area Packing with and without Rotation
Binary Huffman Coding with Unequal Letter Costs Is NP-Complete
Get the few papers worth your time in Robotics every morning.
ReadSearch reads the new papers in your field each day and brings you the few worth reading, each with its key finding in a sentence.