The University of Liverpool has introduced a mobile robotic chemist. The features of this oversized suitcase on wheels with an industrial robot arm on top are described by Burger et al. in the journal Nature (DOI). They are impressive. It navigates through a regular laboratory with a combination of lidar (same as in cars) and touch-feedback with an accuracy of 0.12 mm that it allows humans to work in the same space (lots of industrial robots are lethal and therefore live in cages). The only thing that stops the robot from working is a battery recharge time of three hours but can be easily remedied. Having the same dexterity and mobility as a human the robot can do a lot of mundane things working with ordinary lab equipment like filling vials, capping ,sonicating and operating a GC.
The robot was put to work in water splitting research. The number of variables in the design (photocatalyst type, pH, ionic strength, surfactant yes/no, scavenger yes/no) would result in nearly 100 million individual experiments. Not that the robot or an army of robots would mind but fortunately Bayesian optimization helped to reduce the number to 700 (clocking 2.7 km in 8 days) required to find a promising condition for hydrogen evolution.
But why is building the robot part academic research? This type of robot should be highly attractive to a lot of large commercial analytical laboratories which should be willing to throw a lot of resources into their development, out-competing universities. After all, the Tesla car is being developed in industry and not in academia. In corona times there is an especially urgent need for analytical laboratories running on a massive scale, our lives depend on it. Not mentioned in the article but robots do not shed DNA - is DNA contamination still an issue in PCR? The acknowledgements make clear the robot arm supplier did help in parts of the design and one commercial company called CSols Inc. (a laboratory informatics company) funded part of the research.