Robotic Probe Speeds Materials Measurement

MIT’s Autonomous Robotic Probe Revolutionizes Materials Measurement

MIT researchers have developed a groundbreaking autonomous robotic system that dramatically speeds up the measurement of photoconductance in semiconductor materials. This innovation could accelerate the development of more efficient solar panels and other advanced electronics, addressing a key bottleneck in materials research.

Traditionally, measuring material properties like photoconductance—how well a material responds to light—has been a slow and manual process. This manual approach has hindered the pace of innovation in semiconductor research. The new robotic system, guided by advanced machine-learning models, automates this process, making it faster and more precise.

The system uses a robotic probe to measure photoconductance by making contact with the material at specific points. The machine-learning model, integrated with expert knowledge from materials scientists, selects the optimal contact points to maximize the information gathered. A specialized path planning algorithm then optimizes the probe’s movement between these points, ensuring efficiency and accuracy.

Enhancing Precision and Speed

During a 24-hour test, the robotic probe took over 125 unique measurements per hour, demonstrating greater precision and reliability than other AI-based methods. This high throughput could significantly accelerate the development of solar panels and other semiconductor-based technologies.

"This system is exciting because it enables autonomous, contact-based characterization methods," said Tonio Buonassisi, professor of mechanical engineering at MIT. "Contact is essential for measuring certain material properties, so speed and information maximization are critical."

The robustness of the machine-learning model is further enhanced by integrating domain expertise from chemists and materials scientists. This integration ensures that the robotic system not only operates quickly but also produces high-quality data.

Innovations in Perovskite Research

Since 2018, Buonassisi’s lab has been developing a fully autonomous materials discovery laboratory, with a focus on perovskites—a class of materials with promising applications in solar technology. Previously, the lab developed methods to synthesize and print perovskite materials, as well as imaging-based techniques to determine their properties.

However, the most accurate way to characterize photoconductance is through direct contact with a probe, combined with measuring the material’s electrical response to light. The new robotic system addresses this need by integrating machine learning, robotics, and materials science into a cohesive autonomous platform.

"To operate quickly and accurately, we needed a solution that would produce the best measurements while minimizing the time it takes to run the entire procedure," said Alexander (Aleks) Siemenn, a co-author of the study published in Science Advances.

Adaptive Machine Learning and Path Planning

The robotic system uses a camera to image slides containing perovskite material. Computer vision algorithms divide the image into segments, which are then processed by a neural network model. This model, informed by domain expertise, selects the optimal contact points based on the sample’s shape and material composition.

"Including human experts is important because robots can improve repeatability and precision," said Siemenn. "The model leverages domain knowledge to ensure that the measurements are accurate and reliable."

The path planning algorithm is crucial to the system’s efficiency. It finds the shortest path for the probe to reach the selected contact points, optimizing the measurement process. The neural network’s self-supervised nature allows it to determine optimal contact points directly on a sample image without the need for labeled training data.

"The printed samples are almost like snowflakes—no two are identical," noted Buonassisi. "The machine-learning approach is adaptable to these unique shapes, making it highly effective."

Rich Data for Rapid Results

The researchers tested each component of the system extensively. The neural network model identified better contact points with less computation time than seven other AI-based methods. The path planning algorithm also outperformed other methods, finding shorter paths for the probe.

During a 24-hour autonomous experiment, the robotic system took over 3,000 unique photoconductance measurements, averaging over 125 per hour. This high volume of precise measurements enables researchers to identify hotspots with higher photoconductance and areas of material degradation.

"The ability to gather rich data at fast rates without human guidance opens opportunities to discover and develop new high-performance semiconductors for sustainability applications like solar panels," said Siemenn.

Future Directions

The researchers plan to continue developing the robotic system to create a fully autonomous laboratory for materials discovery. This advancement could revolutionize the way new semiconductor materials are identified and tested, driving progress in renewable energy and electronics.

This work is supported by First Solar, Eni through the MIT Energy Initiative, MathWorks, the University of Toronto’s Acceleration Consortium, the U.S. Department of Energy, and the U.S. National Science Foundation.