Online Hybrid-Belief POMDP with Coupled Semantic-Geometric Models
Tuvy Lemberg, Vadim Indelman · 2025
Read Online Hybrid-Belief POMDP with Coupled Semantic-Geometric Models by Tuvy Lemberg, Vadim Indelman (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-11202.txt.
Description
Lemberg and Indelman help robots plan safely in worlds where what an object is and where it sits are tangled together. A robot's belief about a room mixes discrete labels with continuous poses, and learned models tie the two tightly, making the label space explode. Their 2025 paper plans under that hybrid belief with a partially observable decision framework.
They define safety that respects meaning, not just geometry. The core trick is a reshaped belief that yields fair samples cheaply. Under stated conditions, expected reward and safety odds come out in closed form across all label assignments. Simulations match the accuracy of brute-force sampling over the whole label space, at polynomial rather than exponential cost. The authors target robots in messy real rooms.
Catalog record
| CREATOR | Tuvy Lemberg, Vadim Indelman |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 106 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-11202