Title:

Developing Methods for Assessing the Efficacy and Accuracy of Human and Robotic Operators in Analytical Chemistry

Nathan Kenyon a, Kanwal Jeet a, Tolulope Ogunsanya a, Anyin Li a* a. Department of Chemistry, University of New Hampshire, 23 Academic Way, Durham, NH 03824  Solution preparation is an integral task in chemical analysis and ensuring accurate results during preparation is vital to obtain accurate data and run successful reactions. To ensure accuracy, there must be a way to find points of failure and give feedback to operators. Equally important is the ability to compare operators, to themselves to measure improvement or deterioration, and to compare separate operators whether that comparison be human to human, robot to human or even robot to robot. With increasing automation in every industry, a method of comparing current and future operators is crucial to allow for the development and eventual maintenance of dynamic robot chemists. To form a method of testing, the task of micro pipetting was used as it is ubiquitous in analytical chemistry for standard creation, sample creation, dilution, and serial dilution. Two orthogonal tests to determine operator error were then developed: 1) gravimetric analysis, measuring one volume and using weight to verify absolute volume, and 2) spectroscopic analysis, measuring two volumes and using absorbance to verify absolute concentration. The resulting data was processed to find percent error of pipetting and then an algorithm (found on page 2) was used developed to find minimum assess operator error and to generate a score that was independent of pipette model. With the algorithm, students in a lab course had The the resulting score distribution was: A's: 17 B’s:4 C’s: 3 F’s: 8 Pipette error correction and scoring algorithm: E_n = specified % error of micropipette at volume of instance n N_E = total quantity of instances with an error E R = total number of instances E_a = calculated % error Minimum Operator Error =E_a-∑_(i=o)^R (E_n*N_E ) x = Minimum Operator Error score= 100/(ln⁡(11)) ln⁡(11-x)

Name

Comment*

Comments are viewable only by submitter

Captcha*