Title :
A direct method for checking overlap of two hyperellipsoids
Author :
Gilitschenski, Igor ; Hanebeck, Uwe D.
Author_Institution :
Intell. Sensor-Actuator-Syst. Lab. (ISAS), Karlsruhe Inst. of Technol. (KIT), Karlsruhe, Germany
Abstract :
In this work, we propose a method for checking whether two arbitrary-dimensional hyperellipsoids overlap without making use of any optimization or root-finding methods. This is achieved by formulating an overlap condition as a polynomial root counting problem, which can be solved directly. The addressed challenges involve the inversion of a polynomial matrix using a direct method. The proposed approach extends one of our earlier results, which was restricted to certain combinations of ellipsoids and yields a fixed run-time for a fixed problem dimensionality. Thus, for the first time, an algorithm for checking overlap of arbitrary hyperellipsoids is proposed that can be evaluated in closed form. That is, in the absence of cut-off errors, the proposed method yields an exact result after a finite number of steps.
Keywords :
computational geometry; matrix inversion; polynomials; arbitrary-dimensional hyperellipsoids overlap; cut-off errors; direct method; fixed problem dimensionality; hyperellipsoid overlap checking; polynomial matrix inversion; polynomial root counting problem; Algorithm design and analysis; Collision avoidance; Computational complexity; Ellipsoids; Optimization; Polynomials; Testing; Hyperellipsoid overlap; Leverrier algorithm; Sturm theorem;
Conference_Titel :
Sensor Data Fusion: Trends, Solutions, Applications (SDF), 2014
Conference_Location :
Bonn
DOI :
10.1109/SDF.2014.6954724