Imperial College London
Canvas Category Consultancy : Research : Academic
Imperial College London’s mission is to achieve enduring excellence in research and education in science, engineering, medicine and business for the benefit of society
Assembly Line
National Robotarium accelerates industry development of wind farm robotics
The National Robotarium is supporting the development of new artificial intelligence and control systems that could enable underwater robots to operate autonomously in turbulent seas, potentially revolutionising maintenance and repair tasks for offshore wind turbines.
“Our trials are showing promising results in enabling underwater robots to maintain stable contact with offshore structures in challenging conditions,” said David Morrison, Project Manager at the National Robotarium. “If successful, the technology could transform offshore wind maintenance, reducing fuel consumption of maintenance missions by up to 97% - from 7,000 litres per day to just 200 litres. This could significantly lower both operational costs and the carbon footprint of maintenance.”
Imperial and BASF spinout SOLVE to digitally transform chemical manufacturing
Pandemic-beating drugs could enter production more quickly and agrichemicals such as fertilisers could be produced with fewer toxic raw materials thanks to technology from the new company SOLVE. The spinout has been launched by Imperial and global chemical company BASF under an innovative partnership model, with funding from BASF subsidiary Chemovator in a pre-seed round led by venture capital firm Creator Fund.
It is using innovative chemical processing techniques to build up large sets of data on chemical reactions, which it will use to train machine learning models to rapidly predict the optimal ways to manufacture high-value chemicals. The company is building up experimental data sets using novel techniques in flow chemistry, an advanced form of processing in which reactions are carried out in a continuous flow rather than in batch vessels. The technology is designed to enable chemical companies to scale manufacturing of new chemicals more quickly and to optimise manufacturing processes.
⚗️ Industry consortium to develop modern chemical manufacturing methods
A major consortium led by Imperial and chemical company BASF is to help make chemical manufacturing more efficient, resilient, and sustainable. Imperial will receive £17.8 million from the Engineering & Physical Sciences Research Council (EPSRC) and industry partners under the EPSRC Prosperity Partnership programme in a consortium of organisations from across the chemicals value chain.
“Flow chemistry is inherently more sustainable than batch processing because it makes better use of heat and materials,” said lead investigator Professor Mimi Hii from Imperial’s Department of Chemistry. “It can also provide a powerful tool for automating production and the research and development of more sustainable processes. However, there are technical bottlenecks that are holding back its full implementation. Through this new consortium we will be in a strong position to address these.”
SnAKe: Bayesian Optimization with Pathwise Exploration
Bayesian Optimization is a very effective tool for optimizing expensive black-box functions. Inspired by applications developing and characterizing reaction chemistry using droplet microfluidic reactors, we consider a novel setting where the expense of evaluating the function can increase significantly when making large input changes between iterations. We further assume we are working asynchronously, meaning we have to select new queries before evaluating previous experiments. This paper investigates the problem and introduces ‘Sequential Bayesian Optimization via Adaptive Connecting Samples’ (SnAKe), which provides a solution by considering large batches of queries and preemptively building optimization paths that minimize input costs. We investigate some convergence properties and empirically show that the algorithm is able to achieve regret similar to classical Bayesian Optimization algorithms in both synchronous and asynchronous settings, while reducing input costs significantly. We show the method is robust to the choice of its single hyper-parameter and provide a parameter-free alternative.
Flexible robotic arm put to work with AR
According to Imperial, the flexible arm can twist and turn in all directions, making it customisable for applications in manufacturing, spacecraft maintenance, and injury rehabilitation. In use, people working with the robot would manually bend the arm into the precise shape needed for each task, a level of flexibility made possible by layers of mylar sheets inside, which slide over one another and can lock into place. So far, configuring the robot into specific shapes without guidance has presented challenges.
To enhance the robot’s user-friendliness, researchers at Imperial’s REDS (Robotic manipulation: Engineering, Design, and Science) Lab designed a system for users to see in AR how to configure their robot. Wearing mixed reality smartglasses and through motion tracking cameras, users see templates and designs in front of them superimposed onto their real-world environment. They then adjust the robotic arm until it matches the template, which turns green on successful configuration so that the robot can be locked into place.
Automation for the people: Training a new generation of chemists in data-driven synthesis
Robotic systems that run multiple simultaneous reactions have been widely used in industry since the 1990s to optimize reaction conditions or screen catalysts, for example. Their robotic arms dispense reagents into racks of vials or into plates containing up to 1,536 individual wells, which serve as miniature reaction vessels. These systems are certainly fast, but they still need a lot of tending by human acolytes.
The answer, many researchers believe, is more data—and lots of it. Researchers have previously tried to train machine-learning algorithms by feeding them data from the chemical literature, but this comes with a lot of drawbacks. For starters, much of the information in a published chemistry paper is not in a machine-readable format, and often it is not linked to the underlying raw data. Published chemistry also tends to be highly biased toward conditions that scientists have previously determined to work for a particular reaction (Nature 2019, DOI: 10.1038/s41586-019-1540-5). All too often, chemists don’t quantify details such as a room’s temperature or the exact time that a reaction took to be completed. And chemists have a bad habit of not providing details about reactions that did not produce the outcome they hoped for, leaving an enormous amount of potentially useful information unpublished, further skewing a computer’s training set. “Synthetic chemists in academic labs are not collecting the right data and not reporting it in the right way,” says Benjamin J. Deadman, ROAR’s facility manager.