Abstract
MeshNav3D is a Python-based framework that addresses the lack of reproducible, standardized benchmarking tools for motion planning on uneven 3D terrain represented as triangle meshes. It supports common mesh formats (.obj, .ply) and integrates a range of classical and modern planners, including A*, Dijkstra, Fast Marching, Heat Method, and exact geodesic solvers like MMP. The software features interactive visualization and automated performance evaluation across multiple metrics. Architected for modularity and extensibility, MeshNav3D is openly available and maintained in a public repository, enabling straightforward reuse, comparison, and extension in robotics and autonomous navigation research.
