To appreciate why the contribution matters, it helps to understand how modern optimal planners work. Algorithms such as Rapidly-exploring Random Tree star, or RRT*, introduced by Karaman and Frazzoli in 2011, guarantee asymptotic optimality, meaning that given enough samples the solution converges to the best possible path. The catch lies in the rate of that convergence. Early samples tend to produce long, winding, inefficient routes, and the planner must then sift through an astronomically large configuration space to find better alternatives. Informed sampling techniques addressed this by restricting new samples to a subset of the state space that could conceivably contain an improvement. The most famous example, Informed RRT*, draws samples from an ellipsoidal region whose size shrinks as the current best path cost decreases. Any point outside the ellipse, by the triangle inequality, cannot lie on a path shorter than the current solution, so sampling there is wasted effort.

The Toronto-led team, however, identified a subtle weakness in this widely adopted strategy. The informed subset is only as tight as the current solution path allows. When that path contains redundant detours, unnecessary loops, or tortuous segments, the ellipsoidal informed set remains bloated, and the planner continues to scatter samples across regions that are almost certainly useless. In other words, the quality of the guidance depends on the quality of the incumbent solution, and early solutions are typically poor. The researchers’ prior work proposed the greedy informed set as a remedy: instead of defining the sampling region by the total cost of the current path, it uses the maximum heuristic cost along that path, producing a substantially smaller region that focuses samples where improvements are genuinely plausible.

In the new article, the authors go well beyond the initial proposal by formally characterizing how the greedy informed set behaves within RRT-like planners. They analyze what happens to exploration when sampling is biased so aggressively, a crucial theoretical question because over-constraining the search risks excluding regions that a future, better solution might need to traverse. The analysis demonstrates that greedy sampling can be integrated without sacrificing the properties that make RRT attractive: the planner still finds an initial solution rapidly and still converges asymptotically to optimal paths. This combination of aggressive focus and retained completeness guarantees is the theoretical heart of the paper, and it distinguishes the approach from heuristic shortcuts that speed things up at the cost of optimality assurances.

Building on this foundation, the team presents Greedy RRT*, abbreviated G-RRT*, a bi-directional anytime variant of RRT* that exploits the greedy informed set in practice. The bi-directional design means the planner grows search trees simultaneously from the start and goal configurations, meeting in the middle, a strategy long known to accelerate the discovery of initial solutions. The anytime property means the planner can be stopped at any moment and return the best path found so far, then continue improving that path if given more time. By combining these established design principles with greedy informed sampling, G-RRT* concentrates its computational budget on the most promising pockets of the search space, tightening the sampling region aggressively as soon as any solution exists.

The empirical evaluation is notably broad. The researchers tested G-RRT* on abstract planning benchmarks, which isolate algorithmic behavior from the idiosyncrasies of any particular robot, and on manipulation tasks drawn from the MotionBenchMaker dataset, a tool developed by Chamzas and colleagues to generate standardized, reproducible motion planning benchmarks for robotic manipulation. They also tackled a dual-arm Barrett WAM problem, a genuinely high-dimensional scenario in which two articulated arms must coordinate their movements without colliding with each other or the environment. Across these domains, G-RRT* found initial solutions quickly and converged to optimal paths faster than state-of-the-art sampling-based planners, including informed variants that rely on the conventional ellipsoidal subset.

The practical implications extend across robotics. Manipulation planning for industrial arms, warehouse robots, surgical systems, and space robotics all involve configuration spaces with a dozen or more degrees of freedom, exactly the regime where sampling-based planners dominate and where convergence speed determines whether a robot can plan in real time or must pause awkwardly while computing. Faster convergence to high-quality paths translates directly into robots that respond more quickly, consume less energy executing shorter routes, and can replan on the fly when the environment changes. The authors note that implementations of G-RRT* are publicly available both within the Open Motion Planning Library, the field’s standard open-source planning framework maintained through OMPL, and in VAMP, a vector-accelerated motion planning platform, lowering the barrier for other research groups to adopt and extend the technique.

The study also situates itself within a rich lineage of sampling heuristics. Researchers have long experimented with biasing samples toward free space, toward narrow passages, or along potential fields, and variants such as RRT-Smart, Quick-RRT, and F-RRT* have each proposed tricks for accelerating convergence, often by shortcutting or smoothing the incumbent path. Batch-oriented planners such as Batch Informed Trees and bidirectional asymmetric search methods like Adaptively Informed Trees represent a parallel line of work that fuses sampling with graph-search ideas. The greedy informed set contributes a distinct and principled mechanism to this ecosystem: rather than post-processing paths or restructuring the search, it redefines the geometry of the sampling region itself, using a heuristic bound that remains valid even when the current path is far from optimal. Recent work on vectorized planning, exemplified by the VAMP system from Thomason, Kingston, and Kavraki, has shown that raw computational throughput can be pushed to remarkable heights, and greedy sampling composes naturally with such speedups because it reduces the number of wasted samples rather than the speed of each sample.

Simulations for the study were performed in an Ubuntu 20.04 Docker container on an Intel Core i7-10875H processor with 40 gigabytes of memory, with all planners implemented in C++, and path simplification was evaluated only in separate post-processing experiments so that it would not contaminate the convergence measurements. This methodological care matters in a field where benchmarking practices have historically varied widely, and the authors leveraged the Planner Developer Tools framework, designed for reproducible experiments and statistical analysis of motion planners. The paper’s theoretical proofs, which the authors note benefited from assistance with the formal arguments, establish the asymptotic guarantees that practitioners need before trusting a new sampling bias in safety-relevant applications.

What makes this work resonate beyond its immediate subfield is the way it reframes a familiar trade-off. Exploration versus exploitation is a tension that recurs throughout artificial intelligence, from reinforcement learning to combinatorial search, and motion planning offers an unusually concrete instance of it. The greedy informed set shows that, with the right heuristic characterization, a planner can lean hard toward exploitation, sampling only where improvement is provably possible, without forfeiting the exploratory completeness that guarantees eventual optimality. As robots proliferate in homes, hospitals, factories, and hazardous environments, the demand for planners that deliver near-optimal paths in milliseconds rather than seconds will only intensify. Greedy RRT* offers a mathematically grounded step toward that goal, and its availability in open-source planning libraries means the broader robotics community can put the greedy principle to work immediately.

Subject of Research: Greedy informed sampling for asymptotically optimal sampling-based robot motion planning in high-dimensional state spaces

Article Title: Greedy heuristics for sampling-based motion planning in high-dimensional state spaces

Article References: Kyaw, P. T., Le, A. V., Mohan, R. E., & Kelly, J. (2026). Greedy heuristics for sampling-based motion planning in high-dimensional state spaces. Autonomous Robots, 50(4), Article 44. https://doi.org/10.1007/s10514-026-10269-0

Image Credits: AI Generated

DOI: 10.1007/s10514-026-10269-0

Keywords: sampling-based motion planning, RRT*, greedy heuristics, informed sampling, robot motion planning, high-dimensional state spaces, asymptotic optimality, bidirectional search, robotic manipulation, MotionBenchMaker, OMPL, path optimization

The shape descends directly from a 1928 discovery by the German mathematician Erik Schönhardt, who constructed a deceptively simple object: a triangular prism twisted so that its two triangular ends rotate out of alignment. In this twisted prism, Schönhardt showed, every diagonal that is not part of the boundary lies entirely outside the body. Pick any four of its six vertices and the tetrahedron they span pokes out of the solid. There is, quite literally, no interior tetrahedron compatible with the twisted faces, so the shape cannot be divided into tetrahedra without adding new points. In 1911, N. J. Lennes had already sketched a similar obstruction, but Schönhardt’s six-vertex example became the canonical specimen, and he proved that any simple polyhedron with the same property needs at least six vertices.

For nearly a century, mathematicians refined and extended this three-dimensional curiosity. Bagemihl replaced one twisted edge with a concave curve studded with extra vertices, producing nontriangulable polyhedra with any number of vertices from six upward. Rambau widened the top and bottom triangles into larger polygons, obtaining twisted prisms over any n-gon. Kuperberg produced an even stranger body in which no tetrahedron built from the original vertices contains the body’s midpoint. What all of these examples share is a single common thread: a fixed boundary face that cannot be extended into a simplex lying wholly inside the solid, either because every candidate simplex sprouts an exterior edge or because the combinatorics of the face lattice forbid a compatible decomposition.

The new work, published in the open-access journal Heliyon by Antal Joós, Attila Kovari, and Bálint Nagy, takes the decisive step into the fourth dimension. The authors construct an eight-vertex body they call B: take a regular tetrahedron as the base, place a second tetrahedron parallel to it at height 2, and rotate the top copy by a small angle of negative 0.01 radians within the appropriate coordinate plane. The convex hull of these eight points is a twisted four-dimensional prism. The mathematicians then carve away four carefully chosen four-dimensional simplices whose interiors are pairwise disjoint, and what remains is a genuine 4-polytope whose facets are all tetrahedra, a direct higher-dimensional analogue of Schönhardt’s construction.

The proof of nontriangulability is a small masterpiece of algebraic bookkeeping. The authors introduce a determinant function that measures, for any point in four-dimensional space, which side of a hyperplane it lies on. Exact sign computations show that the four removed simplices are cleanly separated, that the eight designated points are truly vertices, and that a specific interior point q, the average of all eight vertices, survives the carving and sits strictly inside the remaining body. The crux comes from the base facet: in any hypothetical triangulation, that tetrahedral facet must itself be the face of some 4-simplex, and the only candidates are the four simplices obtained by joining the base to one of the four top vertices. Each of these, the determinant tables show, overlaps the interior of one of the removed pieces. No candidate fits. The contradiction is absolute, and the four-dimensional body is certified nontriangulable.

Remarkably, the obstruction is robust. Because the proof rests on strict determinant inequalities, small perturbations of the vertex positions or the twist angle preserve nontriangulability, meaning the example is not a knife-edge accident but a stable geometric phenomenon. The authors leave tantalizing questions open: over exactly what range of rotation angles does the body remain nontriangulable, what affine transformations can the tetrahedral layers tolerate, and could cubes or other shapes replace the tetrahedra in the construction?

Equally important are the dimension-raising lemmas that follow. The team proves that a pyramid erected over any nontriangulable base is itself nontriangulable, that any polytope carrying a nontriangulable facet in a supporting hyperplane inherits the property, and that bipyramids and suitably split bodies extend the same obstruction upward. These lemmas generate infinite families of nontriangulable polytopes in every dimension from four upward. With characteristic rigor, the authors also demonstrate the limits of such inheritance: they exhibit a four-dimensional body that is perfectly triangulable yet whose cross-section is the nontriangulable Schönhardt polyhedron, proving that a bad cross-section alone does not doom a higher-dimensional body. The precise sufficient conditions, distilled into their Proposition 1, specify exactly when a nontriangulable three-dimensional time slice forces the full four-dimensional spacetime body to be nontriangulable.

That spacetime framing is where the mathematics collides with the real world. Many modern engineering and medical problems treat time as a genuine fourth coordinate: a beating heart imaged by 4D CT, airways swelling and contracting through the breathing cycle, coronary arteries reconstructed dynamically for fluid simulations, robots navigating a workspace while an explicit clock ticks. In each case, the natural geometric object is a four-dimensional polyhedral body, and the natural discretization is a mesh of 4-simplices, the four-dimensional analogues of tetrahedra. The new theorems show that when the underlying three-dimensional geometry is sufficiently twisted or nonconvex and the conditions of Proposition 1 hold, simplex-only meshes built from the original sampled vertices are not merely inconvenient but mathematically impossible. Software must insert additional Steiner points or refine the domain, and the new results quantify why that insertion is unavoidable rather than a limitation of current algorithms.

The medical implications are concrete. The inner surface of a heart ventricle is grooved and trabeculated, and if that nonconvexity is strong enough, even a single three-dimensional snapshot can resemble a theoretical twisted polyhedron; stacking the cardiac cycle into a four-dimensional mantle may inherit the meshing obstruction. Bronchial trees, with their branching curvatures and pocketed geometry, present similar risks. In robotics, motion planning through elongated twisted corridors, or coordination of multiple arms whose combined configuration spaces are highly nonconvex, may force planners to add nodes or refine the discretized space in ways the theorems now anticipate. Engineers analyzing deforming fuselages, thermally buckling structures, or pressure-fluctuating pipelines face the same lesson whenever a three-dimensional shape deforms through time in a nontriangulable fashion.

The authors are candid about the computational frontier that remains. Deciding whether an arbitrary polytope is nontriangulable is hard, and even minimum-size triangulations of convex three-dimensional polytopes are known to be NP-hard, so every guaranteed Steiner point carries a cost that can grow with geometric complexity. The new certification for the four-dimensional family, however, is refreshingly checkable: the determinant signs can be evaluated directly from the vertex coordinates. Future work will need adaptive algorithms for optimal Steiner placement, robustness analysis under the noisy vertex data that real medical imaging delivers, and possibly machine-learning heuristics to predict which configurations demand extra points. For now, the message of this paper resonates far beyond the seminar room: nontriangulability is no three-dimensional curiosity. It lives robustly in higher dimensions, it travels through time, and it tells the engineers and clinicians of the four-dimensional data age exactly when their meshes must break the old rules and add new points.

Subject of Research: Higher-dimensional nontriangulable polytopes and their role in spacetime mesh generation

Article Title: Higher-dimensional nontriangulable polytopes: Theory, proofs, and implications in engineering and medical simulation

Article References: Joós, A., Kovari, A., & Nagy, B. (2026). Higher-dimensional nontriangulable polytopes: Theory, proofs, and implications in engineering and medical simulation. Heliyon, 12(15), Article e45442. https://doi.org/10.1016/j.heliyon.2026.e45442

Image Credits: AI Generated

DOI: 10.1016/j.heliyon.2026.e45442

Keywords: nontriangulable polytopes, Schönhardt polyhedron, computational geometry, triangulation, four-dimensional geometry, Steiner points, mesh generation, 4D medical imaging, robot motion planning, combinatorics, polytopes, spacetime meshing