A robot helps peel lettuce, tiny viruses seek out E. coli in drinking water, computers give a hand to air traffic controllers, and a fascinating and highly mobile gene can facilitate communication within the body — and maybe help treat problems in there too. In this week’s coolest scientific discoveries, it takes a village.
What is it? At the University of Cambridge, scientists have created a lettuce-peeling robot.
Why does it matter? Lettuce count the ways! Researchers think the automation of routine processes like peeling vegetables will be increasingly important in a future marked by growing demand for food, less labor availability and a changing climate. And given that highly variable and delicate tasks like peeling lettuce have been one of those things in which humans still easily beat robots hands down, this advance suggests AI has moved past its salad days. Luca Scimeca, who worked on the project, says future roboticists should be able to take a leaf from the Cambridge book: “The computer vision we have developed, which lies at the heart of our lettuce peeling robot, can be applied to many other crops.” Even cauliflower.
How does it work? The researchers call their system a lettuce and stem detection algorithm: Developed in Cambridge’s Machine Intelligence Laboratory, the bot uses suction, delivered through a 3D-printed nozzle, to tear lettuce pieces off in a way that doesn’t damage the stem. The main action here is the “computer vision” the bot relies on to analyze the stem of the lettuce: Researchers wrote an algorithm that was able to locate the center of the lettuce head, allowing it to remove the leaves with minimal damage to the vegetable.
What is it? Publishing their findings in Analyst, food scientists at Cornell University have genetically engineered a bacteriophage — a virus that targets bacteria — to detect E. coli in drinking water.
Why does it matter? Usually contaminated water — or water suspected of contamination — has to be sent to a lab for testing, a process that can take days. But unleashing the phages, as they’re called, gets results much sooner, which will be particularly useful for hard-to-reach spots in the world where water contamination is a persistent problem. Sam Nugen, a Cornell associate professor of food science, said, “These phages can detect their host bacteria in sensitive situations, which means we can provide low-cost bacteria detection assays for field use — like food safety, animal health, bio-threat detection and medical diagnostics.”
How does it work? As viruses that infect bacteria, phages are some of the most abundant organisms on the planet, and their properties have long intrigued scientists: for instance, as a way to forestall the coming antibiotics crisis. At Cornell, researchers zeroed in on one particular phage called T7NLC, which has a gene similar to the one that enables fireflies to glow. When their genetically engineered phage finds an E. coli bacterium, it infects it and causes it to luminesce. If that’s not a reliable rule of thumb yet, it may as well be: Don’t drink water that’s literally glowing.
What is it? In Europe, the MALORCA project — if you squint, that stands for Machine Learning of Speech Recognition Models for Controller Assistance — has developed speech-recognition technology for air traffic controllers, a development that might reduce workload and increase efficiency in air traffic management.
Why does it matter? While so much of the world is going digital and virtual, air traffic control still relies heavily on voice-based communication between the pilot and the tower; software that can simplify communication by converting speech to text, then, has been a goal in this arena. MALORCA’s ambition has been to achieve this not through expensive and labor-intensive manual programming but through machine learning — training computer networks to analyze speech and thereby learn to recognize it, taking into account not just background noise but regional accents and the like.
How does it work? As a release from MALORCA explains, through the heavy analysis of data derived from air navigation service providers. Basically, a lot of listening and learning on the computer’s part: “Machine learning employs statistical techniques that enable computer systems to ‘learn’ and improve their performance on specific tasks over time by exploiting this data, without being explicitly programmed. This will replace much of the manual effort previously required and reduce costs as machine learning of ABSR models makes adaptation to different airports and maintenance cheaper and faster.” Now that the technology has been developed, the next step could be taking the generic algorithm and seeing how it works at specific airports.
What is it? A collaboration between scholars at the University of Wisconsin-Madison and Saudi Arabia’s King Abdullah University of Science and Technology has produced a “solar flow battery,” which combines the abilities of both a solar cell and a battery into one device.
Why does it matter? As we’ve noted, a persistent problem for renewable energy sources like solar is: What happens when the sun doesn’t shine? This is called the intermittency problem and, as UW-Madison notes, it’s an obstacle to widespread solar adoption in Europe and elsewhere. The device produced by this collaboration is able to absorb solar energy when the conditions are right and store it as chemical energy to be used when needed. As the team reported in the journal Chem, it achieves a more than 14 percent return of solar energy to electricity — what Madison professor Song Jin says is “the highest reported number for an integrated solar-electric and battery system.”
How does it work? By the efficiency achieved in marrying a redox flow battery — which stores chemical energy between two components dissolved in liquid, and is already in use in some places to store energy for the electrical grid — to a high-performance solar cell, without any intermediate steps. Jin said, “We only used two conversions — from sunlight to stored chemical energy to electricity. Combining the functions of separated devices into a single device allows us to bypass the intermediate step of electricity generation, which results in a more efficient, compact, and cost-effective approach to utilizing solar energy.”
What is it? Tiny extracellular sacs called microvesicles could function as “biological delivery trucks,” marshalled by doctors and scientists to target tumors, halt an asthma attack or track toxins, according to new findings from Harvard.
Why does it matter? Harvard professor Quan Lu had been conducting research on a gene suspected of causing asthma when his team made an adjacent discovery: They came across another gene, ARRDC1, whose protein “can travel from inside to the surface of a cell to form tiny sacs, called microvesicles, which are subsequently released into the space around a cell,” according to Harvard Public Health magazine: the kind of mobility that could lead to the above-mentioned medical triumphs. Philip Stahl, a professor of cell biology, commented on the discovery, “This is a whole new form of intercellular communication. It’s going to change medicine. We are learning the way cells talk to each other, but we don’t know all the vocabulary yet.”
How does it work? Microvesicles themselves are only a fairly recent subject of scientific pursuit — the ones Lu’s team focused on are 100 nanometers in diameter, or about 1/1,000th the width of a human hair. Though scientists once thought of extracellular vesicles as a kind of biological debris — the leftovers — they’re now coming to understand that they play a crucial, if still somewhat mysterious, role in the communication between cells, carrying proteins, nucleic acids and lipids. Lu thinks microvesicles could potentially be used as a kind of cellular Trojan horse, for instance, targeting tumors with cancer-killing drugs that the body might otherwise reject.