aiqbenchInformații despre AI & calcul cuantic

...

Română · beta. Traducerea a fost făcută automat și este încă în curs de revizuire. Citește pagina în engleză

perspectiva aiqbenchCercetare AI

This research introduces GBPP, a two-stage learning framework that efficiently predicts optimal base poses for mobile manipulators using RGB-D data. By combining low-cost heuristic labeling with targeted simulation refinement, the method enables rapid online selection of safe and reachable stances without heavy task-and-motion optimization. The approach demonstrates practical value in robotics by bridging geometry-aware planning with data-efficient training strategies.

Sursă care acoperăarXiv

Ce este aceastaLectura unui model asupra articolului și a acoperirii conexe, scrisă pentru aiqbench. Nu sunt cuvintele publicației.

GBPP: Grasp-Aware Base Placement Prediction for Robots via Two-Stage Learning

arXiv31 iulie 20262 min de cititArticolul și perspectiva sunt în engleză.

arXiv:2509.11594v3 Announce Type: replace-cross Abstract: GBPP is a fast learning based scorer that selects a robot base pose for grasping from a single RGB-D snapshot. The method uses a two stage curriculum: (1) a simple distance-visibility rule auto-labels a large dataset at low cost; and (2) a smaller set of high fidelity simulation trials refines the model to match true grasp outcomes. A PointNet++ style point cloud encoder with an MLP scores dense grids of candidate poses, enabling rapid online selection without full task-and-motion optimization. In simulation and on a real mobile manipulator, GBPP outperforms proximity and geometry only baselines, choosing safer and more reachable stances and degrading gracefully when wrong. The results offer a practical recipe for data efficient, geometry aware base placement: use inexpensive heuristics for coverage, then calibrate with targeted simulation.

Abstract:GBPP is a fast learning based scorer that selects a robot base pose for grasping from a single RGB-D snapshot. The method uses a two stage curriculum: (1) a simple distance-visibility rule auto-labels a large dataset at low cost; and (2) a smaller set of high fidelity simulation trials refines the model to match true grasp outcomes. A PointNet++ style point cloud encoder with an MLP scores dense grids of candidate poses, enabling rapid online selection without full task-and-motion optimization. In simulation and on a real mobile manipulator, GBPP outperforms proximity and geometry only baselines, choosing safer and more reachable stances and degrading gracefully when wrong. The results offer a practical recipe for data efficient, geometry aware base placement: use inexpensive heuristics for coverage, then calibrate with targeted simulation.

Robotics (cs.RO); Artificial Intelligence (cs.AI)

arXiv:2509.11594 [cs.RO]

(or arXiv:2509.11594v3 [cs.RO] for this version)

Text integral sub licența deschisă a publicației · 142 cuvinte

Citește pe arXiv ↗Întreabă despre acest articol