Markerless Tracking Simulation for Component Shape Decision Support and Its Verification

Abstract

Markerless tracking methods have undergone continuous improvements in terms of the required computing power, tracking stability and reliability. However, accuracy and precision of markerless tracking depends not only on the algorithm used and the environmental conditions, but essentially on the shape of the tracked object. For example, the tracking result is influenced by the component’s dimensions, symmetries, edges, occlusions, and colors.

To support design decisions in order to obtain a component with an optimal shape for good tracking results, a simulation with parameters of the component, the tracking camera, and the environment can be performed during the design phase. However, the projection of simulation results into real environments can only be verified on an experimental basis. In an experimental setup, the position and orientation of a previously designed component are recorded with the help of markerless tracking at predefined positions to assess its tracking performance. An examination of the measurement results confirms the different tracking ability of the regarded objects in this paper. This work presents a method to evaluate the expected tracking performance with regard to the design of a component before it is actually used in a real environment.