This paper focuses on the simultaneous extrinsic-odometric calibration of a mobile robot system equipped with odometric devices and a monocular camera. Most current approaches are based on either optimization or Gaussian filter, which depends on a manually provided initial guess. In this paper, we propose a two-step fully automatic calibration algorithm, which does not require any prior knowledge of the un-calibrated parameters. In the first step, both the odometric parameters and the extrinsic ones are estimated through a non-iterative auto initialization process. In the second step, a joint optimization problem is solved iteratively to obtain a refined calibration result. By exploiting the planar motion constraints of the landmark measurements, our auto initialization method outperforms a comparison approach in robustness against the image noise. Experiments are conducted with data sets collected from both simulation and an autonomous guided vehicle system, which validates the improvement.