Human-like robotic dosing using Machine Learning

Dose-any-Substance INVITE GmbH
Dose-any-Substance INVITE GmbH

robotic dosing process learned by experience

Due to the technologic advances in robotics and automation of the last decades, there are various automated solutions for even complex manual laboratory processes available. Substance dosing is one of the lesser complex processes, which is why there are already many established automation solutions on the market. Despite the advances they offer, we observe, that dosing or weighing-in of substances is still executed manually in most pharmaceutical labs.
One main reason for that is, that many of these automated dosing platforms do not meet the flexible requirements of these labs. The desired dosing amount may be out of range of equipment capabilities, or the dosing system may not be able to dose different kinds of substances or to handle different types of containers. Besides, some dosing systems do not utilize pharmaceutically cleanable reusable equipment and have to be used as one way product.

INVITE is currently developing and testing a flexible dosing system, that fulfills the mentioned requirements. The approach is simple: If a human lab technician is able to dose any substance with simple equipment (e.g. a measuring spoon), we can develop a system that imitates and learns the manual dosing process using modern robotics. The precise handling of the dosing equipment (e.g. a measuring spoon) is being learned through a supervised machine learning algorithm, which determines the optimal motion parameters for precise dosing. Furthermore, the flexibility of robotic manipulators and image recognition technology enable the handling of different containers and cleanable dosing equipment

On October 26, 2021, the first prototype of the proposed dosing system was presented at the Lab of the Future Online Conference.