LLM + Reinforcement Learning for Autonomous Construction: Mobile Crane Control via ChatGPT 5 and Soft Actor-Critic

Forum topic · 小凯 · 2026-10-22

Summary

Researchers Yifei Xiao, Huasheng Lin, and T.Y. Yang propose an autonomous construction method combining large language models and reinforcement learning, published on arXiv (2610.26899, October 2026). The approach uses an enhanced soft actor-critic (SAC) reinforcement learning algorithm to plan actions for a mobile crane based on construction tasks, with robot kinematics and control algorithms ensuring that generated actions are physically executable by the crane. A custom embedding model extracts environment information relevant to user interests, while ChatGPT 5 updates construction progress, provides safety information to the control system, and generates progress reports. The method was validated through multiple full-scale retaining wall construction experiments, demonstrating its feasibility. The work addresses the gap between traditional construction methods and growing global housing demand, illustrating how large language models can support the transition from traditional construction to smart construction by integrating LLM-driven task understanding and reporting with learning-based crane motion planning.

Overview

Research area: Machine Learning Authors: Yifei Xiao, Huasheng Lin, T.Y. Yang Published: 2026-10-22 arXiv: 2610.26899

This paper proposes a large language model and reinforcement learning based autonomous construction method, motivated by the fact that rapid global population growth is making traditional construction methods insufficient to meet housing demand.

Key contributions

Enhanced soft actor-critic (SAC) RL algorithm: plans actions for a mobile crane according to construction tasks.
Robot kinematics and control algorithms: ensure that actions generated by the reinforcement learning algorithm can actually be executed by the crane.
Custom embedding model: extracts information from the construction environment based on user interests.
LLM integration (ChatGPT 5): used to update construction progress, provide safety information to the control system, and generate construction progress reports.

Validation

The proposed method was verified through multiple full-scale retaining wall construction experiments, demonstrating feasibility of the combined LLM + RL approach in real construction scenarios.

Context

The rise of large language models in recent years opens new possibilities for the transformation from traditional construction to smart construction, enabling more intelligent and efficient building processes.

Original abstract (excerpt)

> Due to the rapid growth of the global population, traditional construction methods are gradually unable to meet the current housing demand. The advancement of machine learning algorithms has encouraged researchers to explore their applications in the construction industry, leading to more intelligent and efficient building processes. In particular, the rise of large language models in recent years has provided more possibilities for the transformation from traditional construction to smart construction. In this study, a large language model and reinforcement learning based autonomous construction method is proposed. An enhanced soft actor-critic reinforcement learning algorithm is developed to plan actions for a mobile crane according to construction tasks. Robot kinematics and control alg...

*Auto-collected on 2026-10-22.*