Scientists are striving to find latest semiconductor materials that might boost the efficiency of solar cells and other electronics. However the pace of innovation is bottlenecked by the speed at which researchers can manually measure vital material properties.
A totally autonomous robotic system developed by MIT researchers could speed things up.
Their system utilizes a robotic probe to measure a crucial electrical property often called photoconductance, which is how electrically responsive a fabric is to the presence of sunshine.
The researchers inject materials-science-domain knowledge from human experts into the machine-learning model that guides the robot’s decision making. This permits the robot to discover the very best places to contact a fabric with the probe to realize essentially the most details about its photoconductance, while a specialized planning procedure finds the fastest strategy to move between contact points.
During a 24-hour test, the fully autonomous robotic probe took greater than 125 unique measurements per hour, with more precision and reliability than other artificial intelligence-based methods.
By dramatically increasing the speed at which scientists can characterize vital properties of latest semiconductor materials, this method could spur the event of solar panels that produce more electricity.
“I find this paper to be incredibly exciting since it provides a pathway for autonomous, contact-based characterization methods. Not every vital property of a fabric may be measured in a contactless way. If you want to make contact together with your sample, you would like it to be fast and you should maximize the quantity of data that you simply gain,” says Tonio Buonassisi, professor of mechanical engineering and senior creator of a paper on the autonomous system.
His co-authors include lead creator Alexander (Aleks) Siemenn, a graduate student; postdocs Basita Das and Kangyu Ji; and graduate student Fang Sheng. The work appears today in .
Making contact
Since 2018, researchers in Buonassisi’s laboratory have been working toward a totally autonomous materials discovery laboratory. They’ve recently focused on discovering latest perovskites, that are a category of semiconductor materials utilized in photovoltaics like solar panels.
In prior work, they developed techniques to rapidly synthesize and print unique mixtures of perovskite material. Additionally they designed imaging-based methods to find out some vital material properties.
But photoconductance is most accurately characterised by placing a probe onto the fabric, shining a lightweight, and measuring the electrical response.
“To permit our experimental laboratory to operate as quickly and accurately as possible, we needed to give you an answer that may produce the very best measurements while minimizing the time it takes to run the entire procedure,” says Siemenn.
Doing so required the combination of machine learning, robotics, and material science into one autonomous system.
To start, the robotic system uses its onboard camera to take a picture of a slide with perovskite material printed on it.
Then it uses computer vision to chop that image into segments, that are fed right into a neural network model that has been specially designed to include domain expertise from chemists and materials scientists.
“These robots can improve the repeatability and precision of our operations, but it can be crucial to still have a human within the loop. If we don’t have a very good strategy to implement the wealthy knowledge from these chemical experts into our robots, we should not going to have the ability to find latest materials,” Siemenn adds.
The model uses this domain knowledge to find out the optimal points for the probe to contact based on the form of the sample and its material composition. These contact points are fed right into a path planner that finds essentially the most efficient way for the probe to succeed in all points.
The adaptability of this machine-learning approach is very vital since the printed samples have unique shapes, from circular drops to jellybean-like structures.
“It is nearly like measuring snowflakes — it’s difficult to get two which can be similar,” Buonassisi says.
Once the trail planner finds the shortest path, it sends signals to the robot’s motors, which manipulate the probe and take measurements at each contact point in rapid succession.
Key to the speed of this approach is the self-supervised nature of the neural network model. The model determines optimal contact points directly on a sample image — without the necessity for labeled training data.
The researchers also accelerated the system by enhancing the trail planning procedure. They found that adding a small amount of noise, or randomness, to the algorithm helped it find the shortest path.
“As we progress on this age of autonomous labs, you actually do need all three of those expertise — hardware constructing, software, and an understanding of materials science — coming together into the identical team to have the ability to innovate quickly. And that is a component of the key sauce here,” Buonassisi says.
Wealthy data, rapid results
Once they’d built the system from the bottom up, the researchers tested each component. Their results showed that the neural network model found higher contact points with less computation time than seven other AI-based methods. As well as, the trail planning algorithm consistently found shorter path plans than other methods.
Once they put all of the pieces together to conduct a 24-hour fully autonomous experiment, the robotic system conducted greater than 3,000 unique photoconductance measurements at a rate exceeding 125 per hour.
As well as, the extent of detail provided by this precise measurement approach enabled the researchers to discover hotspots with higher photoconductance in addition to areas of fabric degradation.
“Having the ability to gather such wealthy data that may be captured at such fast rates, without the necessity for human guidance, starts to open up doors to have the ability to find and develop latest high-performance semiconductors, especially for sustainability applications like solar panels,” Siemenn says.
The researchers need to proceed constructing on this robotic system as they strive to create a totally autonomous lab for materials discovery.
This work is supported, partly, 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.
