The discovery of new semiconductor materials—those crucial ingredients that fuel solar panels and modern electronic devices—has long faced a stubborn bottleneck: the painstaking, manual process required to test and understand each material’s behaviors. With every new candidate, scientists have had to make careful, slow measurements to determine how the material responds to light—a property known as photoconductance. It’s technical, tedious work that’s kept innovation moving at a crawl.
Now, a team at MIT has designed a system that changes the game entirely. Instead of humans hovering over delicate samples with probes, imagine a fully autonomous robot—imbued with the savvy of materials scientists and the speed of industrial automation—doing it all on its own, and doing it faster than anyone could hope for. Their robot doesn’t just speed up the process; it multiplies it. In round-the-clock tests, it made more than 3,000 photoconductance readings in a single day, not only outpacing any manual method but delivering fine-grained, repeatable measurements that quickly spotlight promising materials or signs of trouble, like degradation.
What sets this robot apart isn’t just the arm that touches the samples. It’s the clever head: a neural network fed with deep knowledge from materials experts, linked to machine learning algorithms and computer vision. The robot’s camera scans each sample, splitting it visually into segments, then the neural network picks the best spots to touch, maximizing the data returned from each test. Planning software ensures the robot’s path is as swift and efficient as possible—even adding calculated randomness so it adapts better to odd-shaped materials like perovskite “blobs” or “jellybeans” that don’t fit standard molds.
This blend of domain knowledge and AI produces consistently reliable measurements, with the robot identifying “hotspots” of high photoconductance and spotting early signs of deterioration that could affect long-term device performance. Notably, unlike most other robotic systems, this one doesn’t require vast training datasets. It operates in self-supervised mode, learning from its own results and adapting to different material shapes as it goes. When benchmarked against seven leading AI-based testing methods, MIT’s system outshined them all—both in speed and measurement precision.
But the researchers emphasize that human expertise still matters. While the robot handles the repeatable, high-precision tasks, the insights and intuition of scientists are embedded within its software—ensuring that the robot’s decisions reflect years of experience.
Looking ahead, the MIT team wants to link this robotic tester with automated material synthesis and imaging, moving closer to the dream of fully autonomous materials discovery labs. Such systems could revolutionize not just solar panel research but also any industry chasing the next breakthrough in electronic materials—accelerating innovation in sustainable technologies worldwide.
This ambitious project is supported by leading organizations in energy and technology innovation, including First Solar, MathWorks, the University of Toronto, and the U.S. Department of Energy. Explore the story in detail at MIT News: https://news.mit.edu/2025/robotic-probe-quickly-measures-key-properties-new-materials-0704
Diese Website verwendet Cookies.