An AI-driven autonomous experimental workstation for physical/chemical adsorption analysis, developed collaboratively by XTALPI (02228), Sinopec (Shanghai) Research Institute of Petrochemical Technology, and Beijing JWGB Sci. & Tech. Co., Ltd., has officially commenced operations. This system aims to propel chemical research and development through intelligent, automated material characterization, creating a crucial physical platform and embodied intelligence foundation for the material discovery engine of "AI for Science." It marks a significant step towards Physical AI by integrating high-throughput data collection and autonomous experimental decision-making for material characterization into an intelligent "AI + Robotics + Multi-Agent" system within an industrial-scale setting for the first time. This lays the groundwork for establishing an intelligent, self-evolving R&D system in materials science that is perceptive, operable, and capable of evolution. The achievement highlights three core technological features: fully automated, unmanned experimentation; a closed-loop for generating high-quality native data; and a modular, integrated smart hardware-software base. This collaboration transforms the critical, high-frequency task of material characterization from an experience-driven, manual process to a data-driven, AI-powered industrial-grade intelligent system, achieving orders-of-magnitude improvements in throughput, data accuracy, and experimental safety.
Strategic Collaboration Establishes Industry Benchmark
In multi-trillion-dollar industries like petrochemicals, new energy, and environmental protection, precise material characterization is a frequent, essential, yet highly labor-intensive and experience-reliant step in the R&D process. Porous materials, such as zeolites, activated carbon, and metal-organic frameworks, are among the most widely used catalysts and adsorbent carriers in petrochemicals. Their properties, including specific surface area, pore size distribution, and surface acidity, directly determine the efficiency and selectivity of core processes like catalytic cracking, hydrotreating, and gas separation, making them akin to the "chips" of the petrochemical industry. Traditional characterization methods heavily depend on manual operation, resulting in low efficiency and throughput. Furthermore, variability due to subjective experience and operational differences leads to poor data consistency, preventing vast amounts of experimental data from effectively training AI models—a major data bottleneck hindering intelligent new material development. To address this challenge, the three partners have established a collaborative innovation ecosystem encompassing industrial problem definition, high-end instrument development, and AI-driven autonomous experimentation. This transforms an analysis process traditionally reliant on intuition and experience into a quantifiable, repeatable scientific methodology, directly tackling core industry pain points. In this collaboration, Sinopec (Shanghai) Research Institute contributed deep industry expertise and precise requirements for high-throughput, high-precision, and high-safety industrial-grade needs, providing rigorous real-world validation scenarios. JWGB, a leading domestic manufacturer of physical/chemical adsorption instruments, provided the high-precision analytical instrument hardware, ensuring the accuracy and reliability of data generated by the AI system. XTALPI, as a leader in AI for Science, contributed the "AI brain" and robotic automation capabilities. Leveraging its intelligent "AI + Robotics + Multi-Agent" system for closed-loop autonomous operation, it upgrades standalone instruments into "AI material characterization scientists" capable of autonomous workflow planning, data analysis, and decision-making. This deep, synergistic innovation sets a new intelligent benchmark for material characterization in the industry.
AI-Driven Autonomous Experimentation Reshapes Core Paradigms
At its core, this workstation is AI-native, integrating flexible, scalable robotic workstations with intelligent algorithms to unify experimental standards, forming an integrated hardware-software intelligent system. It achieves a fully automated, closed-loop process from sample handling to data analysis, building a "high-quality data foundation" for AI for Science in materials research: • High-Throughput Autonomous Experimentation: The workstation supports intelligent expansion with multi-channel physical/chemical adsorption modules. Under unified AI scheduling, it autonomously allocates tasks, enabling 7x24 unmanned operation, achieving an order-of-magnitude increase in daily sample processing capacity and freeing researchers from repetitive tasks. • Native High-Quality Data Closed Loop: Robotic operation standardizes procedures, fundamentally eliminating human error. Integration with core intelligent algorithms ensures ultra-high data consistency and repeatability, providing high-quality data for AI model training and making every dataset traceable and reusable. • Inherent Safety and Intelligent Analysis: Unmanned operation significantly reduces risks associated with manual tasks like liquid nitrogen handling, enhancing lab safety. The system also automates data analysis, result normalization, and trend modeling, laying the groundwork for the digital and intelligent transformation of materials R&D. This system is developed based on XTALPI's mature, customizable AI autonomous experimentation platform, which boasts over 30 functional modules for flexible configuration and is recognized by more than 300 top global enterprises and research institutions.
Building a Data Foundation to Advance Towards Physical AI
Through full-process digitization and autonomous operation, the "Intelligent Physical/Chemical Adsorption Analysis Autonomous Experimental Workstation" achieves the complete capture and structured accumulation of experimental data for the first time, transforming scattered, ephemeral individual experience into accumulative, reusable data assets. Its significance extends beyond point efficiency gains; it provides high-quality "data fuel" for building AI models that can understand and interact with the physical world. This enables true rational material design, allowing the discovery of structure-property relationships from massive experimental datasets and forming a closed-loop evolutionary capability of "experiment-data-model-prediction." A business representative from Sinopec (Shanghai) Research Institute of Petrochemical Technology stated, "This intelligent workstation introduces AI and autonomous experimentation into the core catalyst R&D process. It not only achieves leaps in efficiency and data precision but also provides a new research paradigm for exploring high-performance, low-energy-consumption catalysts. It supports the industry's green, low-carbon transformation under the 'dual-carbon' goals and is a benchmark practice for intelligent transformation in chemical R&D." The head of Automation Innovation Business at XTALPI stated, "This is a significant breakthrough for AI for Science in the petrochemical sector. We are building an R&D infrastructure that enables AI to continuously generate and digest high-quality data to drive self-iteration. Starting from here, this model will expand from adsorption characterization to broader material R&D scenarios, shifting materials development from 'experience-driven' to 'data-driven and intelligence-driven,' empowering innovation at the source for more industries." A representative from Beijing JWGB Sci. & Tech. Co., Ltd. stated, "This collaboration is a model for the integration of domestic instruments with AI autonomous experimentation technology. We will continue to enhance product intelligence, providing global users with more comprehensive digital-intelligent solutions for material characterization." Looking ahead, XTALPI will continue to deepen its platform technologies in AI for Science, driving materials R&D from empirical exploration towards deeper mechanistic understanding and rational design. It will continue refining its R&D flywheel—centered on vertical AI models, large-scale robotic laboratories, and Multi-Agent systems—within industrial settings, accelerating the comprehensive intelligent upgrade of China's materials R&D sector and injecting sustained momentum into serving the national strategy for scientific and technological strength and high-quality industrial development.
Comments