Learning to Build: Autonomous Robotic Assembly of Stable Structures Without Predefined Plans
Learning to Build: Autonomous Robotic Assembly of Stable Structures Without Predefined Plans
This paper introduces an autonomous robotic assembly framework that constructs stable structures without predefined architectural blueprints. By defining tasks through targets and obstacles, the system adapts to environmental uncertainty using a reinforcement learning policy trained with deep Q-learning and successor features. Evaluated on a benchmark of 15 discrete 2D block construction tasks in a real-world closed-loop setup, the approach demonstrates feasibility and robustness against construction noise, suggesting a promising direction for adaptable robotic construction.
The authors present a reinforcement learning framework for autonomous assembly of stable structures without predefined blueprints, evaluated on 15 discrete 2D tasks in a real-world closed-loop setup; results indicate feasibility under construction noise but do not claim general deployment readiness.
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.
- Control
- Reported autonomousOpen source S2
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
arXiv:2602.23934v3 Announce Type: replace Abstract: This paper presents a novel autonomous robotic assembly framework for constructing stable structures without relying on predefined architectural blueprints. Instead of following fixed plans, construction tasks are defined through targets and obstacles, allowing the system to adapt more flexibly to environmental uncertainty and variations during the building process.
Open source S2
Source：arXiv Robotics — research abstracts · arxiv.org