Anticipating Application of Machine Learning Techniques for Effective Synthesis of Straight-Line Crank Rocker

Planar straight-line mechanisms recognized by associated name of the creator are very few and have periodic appearance in last three centuries. Barring a few the evidence of involvement of concepts like the Inflection Circle or Cubic of Stationary Curvature in their development is not available. Presently use of these concepts is avoided due to involved mathematical treatment given to explain them. In this paper these concepts are demonstrated without mathematical jargon for the multidisciplinary teams comprising of data scientists and present day kinematicians. Possibility of using techniques of Machine Learning domain with compact, easy to generate, theme-based databases is explored. Work done so far for completing the numeric modeling in synthesis of straight-line crank rockers is explained. So far Inflection Circle is the theme for creation of Generated Database. Reasons for not creating thematic database using Cubic of Stationary Curvature are explored. For better visualization of the spread of data points of the Generated Database, a surface is plotted. Possibility of developing a novel data-driven model using Machine Learning techniques is contemplated as an alternate for creation and use of a thematic database jointly using both these concepts. Recent published literature about the individual Machine Learning technique (Deep learning NNs, Reinforcement learning and Clustering) for linkage synthesis is explored for their suitability in creating a validated inferential (input output based) representation of straight line crank rockers. This paper explores alignment of the classical synthesis problem with the tools of Industry 4.0 paradigm.