Retrieval-Augmented Robots via Retrieve-Reason-Act CCS: Information systems Language models CCS: Computing methodologies Machine learning CCS: Computing methodologies Robotic planning
Izat Temiraliev, Diji Yang, Yi Zhang · 2026
Read Retrieval-Augmented Robots via Retrieve-Reason-Act CCS: Information systems Language models CCS: Computing methodologies Machine learning CCS: Computing methodologies Robotic planning by Izat Temiraliev, Diji Yang, Yi Zhang (2026) 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-2603-02688.txt.
Description
Izat Temiraliev, Diji Yang, and Yi Zhang argue that robots should look things up before they act. Their 2026 paper defines Retrieval-Augmented Robotics for cases where a robot gets no demonstrations at all. A robot facing a complex furniture kit lacks the exact steps, and its built-in knowledge cannot fill that gap.
Their system loops through three steps: retrieve, reason, and act. The robot searches visual manuals, matches flat diagrams to real three-dimensional parts, and then builds an action plan. On a hard long-horizon assembly test, planning grounded in retrieved manuals beat both unaided reasoning and example-based retrieval. The work extends search from answering questions to driving physical action.
Catalog record
| CREATOR | Izat Temiraliev, Diji Yang, Yi Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 47235 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.02688