AI sorting machine seeks to bolster TCM industry

A Tianjin-based research center has rolled out the latest generation of an artificial intelligence-powered sorting machine for traditional Chinese medicinal herbs, replacing subjective manual inspection with digital quality assessment in a move aimed at modernizing the centuries-old industry.

The new system, developed by the National Modern TCM Innovation Center, combines X-ray imaging, hyperspectral analysis and machine vision to detect hidden defects, measure chemical composition and automatically grade raw herbal materials.

The fourth-generation machine is already being used at pharmaceutical facilities in Tianjin and provinces of Jilin, Jiangsu and Guangdong, where it helps manufacturers identify counterfeit or damaged herbs, remove impurities and sort raw materials according to quality standards.

"Herbs supplied by different farmers can vary significantly in quality," said Liu Changqing, senior manager of the center's green intelligent manufacturing department.

"The machine grades every batch before production begins, helping manufacturers ensure the consistency of finished medicines. That has been the central goal of our research."

Traditionally, medicinal herbs were sorted through a combination of manual picking, air separation and visual inspection. The labor-intensive process depended heavily on workers' experience, making it difficult to detect internal defects or maintain consistent standards.

The new system addresses those limitations by combining three types of data.

X-rays examine the internal structure of herbs without damaging them, revealing hidden problems such as rot or hollow cavities. Hyperspectral sensors analyze chemical composition and estimate the levels of active medicinal compounds, while high-resolution cameras inspect surfaces for mold, dirt and other contaminants.

"Each technology captures different information," Liu said. "When the data is cross-checked, the error rate can be greatly reduced."

The research team spent five years developing four generations of the machine. The first prototype was unveiled in 2020, while the latest version entered mass production earlier this year.

So far, researchers have developed digital identification models for more than 50 commonly used medicinal herbs. But the challenge remains substantial.

China uses around 1,000 major medicinal materials, many of which exist in multiple processed forms. Ginger alone, for example, can be processed into six products, including fresh ginger, dried ginger and ginger peel, each requiring large amounts of training data to build accurate algorithms.

The team has collected around 400,000 samples and is currently digitizing about a dozen new herb varieties each year.

"We hope to gradually build digital databases covering all mainstream medicinal herbs," Liu said."We also develop customized algorithms tailored to individual manufacturers and their production processes."

Researchers said the technology could also strengthen the international competitiveness of Chinese herbal products by providing standardized and traceable quality-control records.

According to customs data, China exported 255,000 metric tons of raw herbal materials to 196 countries and regions in 2025. Industry experts said overseas buyers increasingly prefer objective testing data and traceable records over assessments that rely solely on human experience.

During this year's National People's Congress, Zhang Boli, an academician of the Chinese Academy of Engineering and chief technical adviser to the center, said AI could help modernize the TCM industry, but should complement rather than replace professional expertise.

"If Chinese medicine is to advance internationally, standardized digital quality control must become the foundation, rather than relying solely on the experience of veteran pharmacists," Zhang said.

Beyond intelligent sorting hardware, the National Modern TCM Innovation Center has built a full R&D pipeline covering lab trials, pilot manufacturing and large-scale commercial rollout.

Wang Ping, general manager of the innovation center, said the facility connects every stage from lab experiments to mass manufacturing via continuous pilot production lines.

"The quality of authentic medicinal herbs directly shapes how well finished treatments work," Wang said. "Our integrated pilot platform combines extraction, formulation and full-process quality testing under one roof. This significantly lowers costs and technical barriers for small and medium drug firms to upgrade their factories and develop new herbal medicines."

Meng Jiarui, Zhong Aiyamei and Wei Lan contributed to this story.