Reverse engineering (RE) in mechanical studies has many aspects today, and in robotics, it has even more directions and possibilities of study. Such aspects can be included in mechanical studies, hardware, IT (code and control) studies, performance studies, defect and wear studies, identification of mathematical models, systems analysis and their replication studies, and even for educational purposes, etc. A general definition of RE can be considered a return to the initial data of a project, with the information gathered up to the creation of a product, in order to make improvements to it or to understand it more fully, throughout the entire technological development. This article is a study in which we try to improve the kinematic performance of a serial robot by optimizing the initial values of the parameters (the dimensions of the robotic arms) and to understand more deeply the influence of different kinematic optimization methods for workspaces criteria (displacement workspace, velocity workspace, work tool space) or for the robot's dexterity. This research presents results obtained by comparing classical, established methods with methods using genetic algorithms (GA), methods using digital twins (DT), methods using artificial intelligence (AI), or methods using criteria from the neutrosophic theory of mathematical logic.