Experiment planner for strategic experimentation with an automated chemistry workstation

Abstract

An automation system capable of modifying its actions based on experimental data will serve as a useful assistant to the scientist. In order to achieve flexibility in experimentation, we have developed a software system for stating experimental protocols that involve decision-making based on experimental data. These decision-making capabilities enable strategic experimentation, where experiments are altered or terminated based on prior or newly acquired data. An experiment planner provides for experimentation in an automated chemistry workstation using the simplex algorithm or a decision-tree programming approach. The composite modified simplex algorithm is used for optimization experiments. Simplex experiments are planned through a series of menus. The simplex algorithm can be altered in sophistication by turning features on or off in software. The simplex algorithm is well-suited for optimizations but provides no general decision-making capabilities. A general decision-tree programming module enables the composition of diverse experimental protocols which capture the logic of experimentation. Experimental protocols written with a text editor are converted by an interpreter to a set of procedural events. Experimental protocols can be composed that execute incisive actions based on data from ongoing experiments. Protocols have been written for automatically matching samples with analytical instruments and performing an automated dye titration. Resource consumption cannot be predicted in advance in closed-loop experimentation, thus a static resource manager is used with each simplex cycle and a dynamic resource manager is used in decision-tree experimentation. The resource managers and strategic decision-making features empower the workstation to function autonomously in pursuit of the research objectives stated by the scientist.