Chemists have devised a new system to help robots and humans collaborate more efficiently in labs by recognising the strengths and weaknesses of each.
Automation is a growing feature of chemistry labs, taking over many of the repetitive, complex and time-consuming tasks required to run experiments and measure results. In 鈥渟elf-driving laboratories鈥�, robots pour, mix, distil and titrate chemicals, while an artificial intelligence pores through data and highlights interesting experimental avenues.
Al谩n Aspuru-Guzik at the University of Toronto in Canada and his colleagues suggest that a hybrid approach in which humans and robots work collaboratively could be more efficient, with humans stepping in to handle processes that robots find difficult, and vice versa.
Advertisement
To that end, the team has created a complex metric called RouteScore that determines the overall cost of different approaches to synthesise a specific molecule. There can be many routes from raw ingredients to a target molecule, and the system calculates the cost of each, taking into account the time required for the various processes by humans and robots, the costs of materials, the amount of equipment that needs to be cleaned, how much material needs to be disposed of and the cost of machines and labour.
The team ran two experiments. In one, the tool calculated the most efficient way to synthesise a known molecule. In the other, it simulated a self-driving laboratory to find which of more than 3000 possible molecules would be easiest to synthesise. In both cases, the tool was able to identify the optimum process, according to the fluctuating costs of labour and materials.
at the University of Liverpool, UK, says chemists need to be careful of 鈥渁utomating for automating鈥檚 sake鈥�, because it isn鈥檛 always faster or more efficient. 鈥淎 lot of new people coming into automation fall into this trap, and they feel that they have to automate everything, every step of the reaction,鈥� he says.
鈥淲hat I think this will potentially give people is the best balance between what you should automate and what you should do in the lab yourself. One is never 100 per cent better than the other, and the hybrid of the two could actually be more more fruitful.鈥�
at the Massachusetts Institute of Technology says full automation is a goal that should still be pursued, particularly in processes involving very hazardous substances, but that robots and humans will always have different strengths. 鈥淪o it鈥檚 important to figure out how best to integrate the two,鈥� he says. 鈥淪teps towards quantifying their trade-offs are valuable for the field.鈥�
ACS Central Science
