3D sand printing is a key direction for the casting industry toward intelligent manufacturing. However, precisely locating the optimal solution within a high-dimensional, nonlinear parameter space has become a critical bottleneck constraining the high-quality development of 3D sand printing. To address this challenge, the group led by Associate Professor Hu Songtao from School of Mechanical Engineering, Shanghai Jiao Tong University has developed an intelligent experimental system, Ex3DSP. By adopting a “do-while-think” self-evolving experimentation, Ex3DSP achieves a paradigm shift from empirical trial-and-error towards precise targeting. The findings are published in National Science Review, titled “Self-evolving experimental platform for 3D sand printing”.

Ex3DSP integrates an automation-based “capable body” to support standardized experimental data production, and an AI-driven “thinking brain” to enable precise experimental data analysis, achieving the closed loop of “design—fabrication—characterization—analysis”. Remarkably, only 60 experiments were required to map the Pareto front of a three-objective, three-variable optimization problem, reducing experimental effort by three orders of magnitude compared with exhaustive testing. The mechanical strength, gas evolution, and permeability of the sand molds were enhanced by 103.5%, 45.3%, and 25.3%, respectively, ultimately yielding a 385% enhancement in casting performance.

Led by Hu Songtao, the unmanned experimental system group focuses on tackling high-dimensional nonlinear experimental challenges. Targeting diverse application scenarios—including biomedicine, integrated circuits, and additive manufacturing—the group develops autonomous experimental technologies and scientific instrumentation, encompassing robotics, microfluidics, multimodal measurement, artificial intelligence, simulation computing, and agents. Their work has been published in journals such as Science Advances, Nature Communications, and National Science Review, and has been widely applied in national laboratories, large-scale scientific facilities, and industrial enterprises. Their achievements have been featured on CCTV News Broadcast, Xinhua Net, and People’s Daily Online.

Paper Link: https://doi.org/10.1093/nsr/nwag352