Evaluation of dimensional analysis in robotics

Abstract

Robot programs typically manage a substantial amount of dimensioned data. However, existing robot programming tools do not directly support the description and manipulation of dimensioned quantities. A new data type has been proposed for managing dimensioned data in robot software. The design removes much of the hard work of using and ensuring correctness in dimensional data and allows mixing of units using dimensional analysis. An initial implementation has been created by extending the Python interpreter. This paper evaluates the proposal for dimensioned data in robotic systems, using established qualitative evaluation criteria for programming languages, based on the design and implementation, human factors, software engineering, and the application domain. In addition the paper shows that the method can be formalised, and that it is able to support coordinate systems. A set of unit tests is reported for the implementation. Together these evaluations show the proposed dimensioned data system to be implementable, self-consistent, and useful in robotics. The proposed design improves robot programming where dimensioned data is involved.