Code Generation from Declarative Models of Robotics Solvers
Marco Frigerio, Enea Scioni, Pawel Pazderski, Herman Bruyninckx · 2019 Third IEEE International Conference on Robotic Computing (IRC) · 2019
This short paper describes a development methodology addressing the limitations of traditional robot kinematics and dynamics software libraries. Specifically, the implementation choices which hinder the integration of the libraries within user code based on different assumptions; for example, choices about the mathematical formalism or the digital data types. Code generation based on declarative and semantically unambiguous specification is proposed as a more flexible development approach, which allows to configure the generated API and the concrete implementation choices. The paper also introduces a prototype tool that we developed to investigate our research hypothesis, and discusses some of the current challenges.