MiCU: End-to-End Smart Home Command Understanding with Large Language Model

2026-05-31

AI Summary

This work addresses the challenge of accurately interpreting ambiguous or misaligned natural language commands in smart home systems, such as “make the bedroom more comfortable.” To this end, the authors propose MiCU, a domain-specific large language model tailored for smart home environments. MiCU leverages automatically synthesized logs for training data, incorporates curriculum learning to inject domain knowledge, and employs cold-start pretraining combined with rule-based reinforcement learning to enhance reasoning capabilities. A novel single-token compression technique for device descriptions is introduced, substantially reducing inference overhead and enabling efficient processing of long inputs. Experimental results demonstrate that MiCU achieves an average accuracy improvement of 20.01% across all device categories. Upon deployment in the Xiaomi Home App, it attained 1.7 million daily active users, reduced user correction rates by 1.57%, and increased human review accuracy by 32.05%.

Abstract

Command understanding systems in smart home ecosystems can automate device control and substantially improve user experience. However, while they perform well on precise utterances (e.g., "turn on the bedroom light"), they struggle with ambiguous or misaligned commands (e.g., "make the bedroom cozy"). Large language models (LLMs) generalize well across various domains and can outperform traditional rule-based systems on such tasks, but their effectiveness is often constrained by scarce domain-specific data, insufficient task-specific adaptation, and high computational costs. In this paper, we propose an automated training data synthesis workflow using user logs and LLMs; then we build MiCU, a domain-specific LLM that excels at command understanding. Specifically, we employ curriculum learning to inject domain knowledge into the base LLM, then we enhance its reasoning ability via cold-start training combined with reinforcement learning (RL) guided by domain-specific thinking rules. Additionally, we introduce a token compression technique that condenses device description into a single special token, substantially reducing inference overhead and enabling \model-fast, an efficient variant optimized for long inputs. Extensive experiments show that MiCU significantly outperforms baselines, with an average accuracy gain of 20.01% across all device categories. We have deployed MiCU in the Xiaomi Home app, receiving approximately 1.7 million page views per day. Production evaluations show that MiCU reduces user correction rate by 1.57% and increases human audited accuracy by 32.05%. Our data and code are available at https://github.com/xiaomi-research/iot_spec_llm

Problem

• command understanding
• smart home
• ambiguous commands
• large language models
• natural language understanding

Innovation

• data synthesis
• curriculum learning
• reinforcement learning
• token compression
• domain-specific LLM