Embedding Large Language Models into Flow Controls: An Agentic Framework for Adaptive and Trustworthy Automated Cooking

2026-08-05

AI Summary

This work addresses the limitations of conventional automated cooking systems, which rely on rigid procedural scripts and lack personalization and interpretability, as well as the opacity and unreliability of existing large-model-based approaches. The authors propose a multi-agent collaborative framework that translates user intent into structured recipes, explicit workflow control, and executable Python code. Trustworthy automation is achieved through a three-stage pipeline: offline generation, online closed-loop execution, and post-execution adaptation. By embedding large language models within the workflow controller, the system enables verifiable, intervenable, and longitudinally personalized cooking tasks. Experiments on a real-world platform demonstrate high task completion rates across diverse scenarios, transparent execution logic, and effective anomaly handling.

Abstract

Automated cooking robots have traditionally relied on predefined procedures and rule-based control, ensuring stable execution but offering limited personalization, whereas recent large-model approaches support natural language interaction but often suffer from opaque decision making and unreliable execution in real kitchens. To address this challenge, this paper proposes an agentic framework that systematically decomposes personalized cooking requirements into structured and verifiable control programs rather than directly mapping language to actions. Multiple AI agents collaboratively transform user intents into canonical recipes, workflow programs with explicit flow control, and executable Python code grounded in an atomic action library. The system consists of three tightly coupled stages: offline recipe-to-code generation through multiple agents, online closed-loop execution with supervisory intervention enabled by multimodal perception, and post-run adaptation that updates user preference models for long-term personalization. Real-world experiments on a physical cooking platform demonstrate that the proposed framework achieves reliable task completion, transparent execution logic, and effective anomaly handling across diverse personalized scenarios, validating its practicality for trustworthy automated cooking in real environments.

Problem

• automated cooking
• personalization
• trustworthy AI
• large language models
• flow control

Innovation

• agentic framework
• flow control
• large language models
• automated cooking
• multimodal perception