Abstract—This paper presents a method to estimate the two-dimensional mount parameters (offset and rotation) of a bearing only sensor on a mobile robot. It is a continuation of the work done by D. Scaramuzza on auto-calibration of omnidirectional cameras for mobile robots. The system model of the robot is split into two subsystems with reduced complexity, which makes it possible to detect incorrect estimates, as well as to reduce the parameter search space to one dimension for each subsystem. Multiple hypothesis extended kalman-filters (MHEKF) are used to cover a part of the search dimension and the parameter space for a validation parameter. Based on this approach, incorrect estimates can be detected automatically which is a requirement to use calibration methods in a fully automated setting.

Index Terms—omnidirectional camera, calibration, kalman-filter

Cite: Max Fischer, Thimo Langbehn, and Volker Sommer, "Reliable Automatic Calibration of Omni-Cams with Multi-Hypothesis Extended Kalman Filters," Jounal of Automation and Control Engineering, Vol. 2, No. 4, pp. 422-427, December, 2014. doi: 10.12720/joace.2.4.422-427