Chemist Mathijs Mabesoone is working with his colleagues from the Big Chemistry Consortium on a fundamentally different way of doing materials research: the self-learning robot laboratory. By using AI and automation intelligently, the consortium wants to radically accelerate innovation. What does that mean for the laboratory of the future? And who will be turning the knobs?
Many everyday products are complex mixtures, so-called formulations. Their properties are created by subtle interactions between molecules. If a formulation needs to be adjusted – for example due to changing regulations or consumer wishes – we rarely go back to the drawing board. “Take shampoo. If it needs to foam more, we add a molecule,” says Mathijs Mabesoone, group leader of physical-organic chemistry at the Big Chemistry Consortium. “We rarely redesign these mixtures from scratch.” The result: increasingly complex and difficult to understand compositions, which are not always sustainable or efficient.
To change this, Mabesoone is working on a revolutionary robotics lab in which AI and robotization fundamentally change chemical research. The goal is to redesign formulations from the bottom up, based on molecular interactions and a solid scientific foundation. This requires a fully automated lab that generates data itself – day and night, without manual work. This data feeds an artificial intelligence that learns to predict which molecules work together, and why. “A huge benefit is that we can now predict much better which experiments are really useful,” says Mabesoone.
Robots don't take breaks
The scale on which the robotic laboratory operates enables fundamental research through efficiency and time savings. An example is the fully automated analysis of soap. “Traditionally, such an analysis would easily take an analyst a whole day,” says Mabesoone. “Our robot does ten to twenty a day.” The robot works completely independently: it determines the correct dilutions, performs measurements, analyses the results and decides which follow-up steps are required. “And this process continues 24 hours a day, even at night and during the weekend.” Smart automation enables a major efficiency boost. In addition, this new technology creates enormous data sources and space for radical innovation.
However, the robots also encounter unexpected limitations. “Some actions that are very simple for people turn out to be extremely difficult for robots,” says Mabesoone. “Take weighing powders. Liquids can be automated well, but powders are a different story. Even an everyday ingredient like sugar has all kinds of variants – from powdered to crystal and caster sugar – that each behave differently. The robot lab must be able to deal with all those different manifestations of such a solid substance. And it is not that easy to solve that.”
Gut feeling
Robots are lightning fast, precise and tireless, but they lack something fundamental: intuition. “They are extremely good at collecting large amounts of data in a standardized way, or performing repetitive actions,” says Mabesoone. “But they have no gut feeling.” And that is precisely what is crucial. “There are countless examples where intuition was leading. Take the discovery of Teflon: an unexpected reaction during an experiment suddenly produced a material that did not stick to anything. That was unplanned, misunderstood. And it led to a completely new field of research. That is why I think that for fundamental research, where the question is not yet clear, human insight remains indispensable.”
The laboratory of the future
Automated labs make chemical research faster, bigger and more systematic. “Small labs working on general research questions can’t compete with that,” says Mabesoone. In line with that lies the ultimate ambition: a fully autonomous lab, driven by artificial intelligence. A so-called dark laboratory, where no human is involved anymore. “That is the final stage of this project, but we are not there yet.”
“Computers and robots can measure and automate a lot, but a critical pair of eyes remains essential,” says Mabesoone. “The analyst of the future is less hands-on, but monitors the overview and quality of the process.” Education must also move along with this. “We still train chemists too little in programming and data thinking. Within the Big Chemistry Consortium, we are strongly committed to this. We inspire students by having them work on large projects. This allows them to see the impact of their work more quickly. If we really want to train chemists who are ready for the future, we must structurally anchor data and automation in the curriculum.”
Radically different
For Mabesoone, the goal of Big Chemistry is clear. “I hope that with this program we can show that by embracing AI, data and automation, we can make new molecules and materials possible – radically different from what we know now. We want to prove that this approach works, and that we can also take the doubters with us. Because all data, all measurements are valuable – provided you record them properly and share them with the community. If we can achieve that, knowledge will not remain in a dusty cupboard, but each experiment will build on the previous one.”
Want to know more? Mathijs Mabesoone is one of the speakers at the Lab of the Future seminar on Thursday 25 September during the LabNL trade fair. Discover which technologies are making a difference now and in the future, and see how the future of laboratory innovation is taking shape.
Nederlands