VerifLM: Physically Verified Image-Edit-Guided Goal Pose Generation for Robotic Manipulation
VerifLM: Physically Verified Image-Edit-Guided Goal Pose Generation for Robotic Manipulation
This paper introduces VerifLM, a framework that translates image-edited goal configurations into physically verified 6-DoF object poses for robotic manipulation. The system reconstructs metric object geometry from RGB-D observations without predefined meshes or physical parameters, then uses a vision-language model to generate axis-wise sampling constraints and a task-specific validation program. Candidate poses are evaluated in parallel-world physics simulations with varying geometry and physics, and the VLM selects the most task-consistent candidate. Evaluated on nine manipulation tasks, VerifLM achieves the highest success rate on eight of them, showing the largest performance gain for goals requiring precise orientation and contact relationships.
VerifLM converts image-edited goal configurations into physically verified 6-DoF object poses for planning by reconstructing metric geometry from RGB-D observations and evaluating candidates across parallel-world physics simulations, achieving the highest task success on eight of nine evaluated manipulation tasks.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
h varying physics and geometry. Parallel-world validation evaluates candidates across geometry--physics variations and ranks them by the number of worlds in which they satisfy the validation program. We evaluate VerifLM on nine manipulation tasks. VerifLM achieves the highest task success on eight of nine tasks, with the largest performance gain on goals requiring precise orientation and contact relationships.
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org