Design of Ingredient Recognition and Automatic Feeding System for Frying Robot Based on Machine Vision

DOI: https://doi.org/10.62517/jike.202504306

Author(s)

Zhanqiang Liu1, Liang Dong2, Jianling Zhao1,*, Jianguo Wang3

Affiliation(s)

1Shandong Hehe Information Technology Co., Ltd., Jinan, China 2Jinan BOSAI Network Technology Co., Ltd., Jinan, China 3Shandong Tongyun Network Technology Co., Ltd., Jinan, China *Corresponding Author

Abstract

In order to improve the accuracy of food ingredient recognition and the synergy of feeding control in stir-frying automation equipment, a machine vision-based intelligent stir-frying system is designed to construct a hierarchical architecture consisting of visual recognition, control decision-making, and multi-axis actuator units. A high-resolution industrial camera combined with MobileNetV2 network is used for image feature extraction and classification, and control labels and scale parameters are output for the execution system to schedule the feeding path. The path planning and closed-loop control is realized by integrating the STM32F407 master chip and PD control strategy. Experiments show that the system has a recognition accuracy of 93.6% under the conditions of multiple types of ingredients, an average feeding path error of ±1.8 mm, and the response delay is controlled within 842 ms, which verifies the validity and practicability of the synergistic mechanism of visual recognition and automatic control.

Keywords

Machine Vision; Intelligent Stir-Fry Machine; Ingredient Recognition; Automatic Feeding

References

[1] Sha, F., Meng, J., Zheng, X., & Jiang, Y. (2025). Sustainability Under Fire: How China-US Tensions Impact Corporate ESG Performance?. Finance Research Letters, 107882. [2] Hu, X. (2025). Learning to Animate: Few-Shot Neural Editors for 3D SMEs. [3] Hu, X. (2025). Low-Cost 3D Authoring via Guided Diffusion in GUI-Driven Pipeline.