Positive dimensional parametric polynomial systems, connectivity queries and applications in robotics (Q2674013)

This work considers parametric systems of polynomial equalities and inequalities, the positive dimensional semialgebraic sets defined by them and how these depend on the values of the parameters. In particular, it focuses on the number of connected components that the semialgebraic set has for a given point in parameter space. It proposes algorithms to, on the one hand, decide the different possible numbers of connected components throughout the parameter space and provide selected points in that space for which each number of connected components is attained. On the other hand it deals with the construction of a connected path that brings any point in parameter space to one of the selected ones, such that the path stays in one of the connected regions of parameter space yielding constant number of connected components in the corresponding semialgebraic set. Such semialgebraic sets are of interest in real world applications, as illustrated here with the example of kinematic design of a particular class of robots. The algorithms provided solve the former problems under certain regularity conditions on the ideals generated by the polynomials defining the system and the algebraic sets defined by these same polynomials, such as radicality of certain ideals, equidimensionality and codimension of the sets defined by them, and sometimes smoothness or codimensionality of the singular locus large enough. This work extends previous results in [\textit{J. Capco} et al., in: Proceedings of the 45th international symposium on symbolic and algebraic computation, ISSAC '20. New York, NY: Association for Computing Machinery (ACM). 62--69 (2020; Zbl 1486.14075)] by relaxing the regularity conditions on the polynomial systems and introducing a new complexity analysis and a new way of reducing connectivity queries in semialgebraic sets to closed bounded ones. The algebraic and geometric tools used include, for example, Hardt's Theorem and semialgebraic trivialization, Sard's Theorem, the Jacobian criterion and Thom's isotopy Lemma. The computational tools used include elimination and the critical point method. As a main step of the strategy, the problem is reduced to computing roadmaps of closed and bounded semialgebraic sets. The algorithms are implemented in Maple, and a full description and complexity analysis is provided. The singularity-free space of the \textit{UR-series} robot is analyzed by means of the tools and algorithms proposed: the parametric system here is given by the determinant of the Jacobian matrix, which is a polynomial in four variables involving four parameters defining the kinematic singularities of the robot.