Ballistocardiogram (BCG) is the repetitive body movements related to cardiac cycle. By using non-invasive sensor embedded in a car seat, the sensor reads the vibration caused by the BCG as minute voltage at any time while driving, making it possible to measure heartbeats without being restrained and to obtain the occupant's biological information. However, BCG motion measurement while driving is difficult due to the incorporation of road noise whose magnitude is usually larger than the undisturbed BCG signal. In this study, we propose a method to extract a BCG signal from a mixture of the BCG and road noise contained in a piezoelectric film sensor using online auxiliary-function-based independent vector analysis with iterative source steering (online-AuxIVA-ISS). A device simulating BCG was placed on a seating surface, while two sensors were embedded in a seat cushion. A shaker was used to reproduce road noise disturbance on estimating the BCG. Data from the two sensors were separated into two independent signals using the online-AuxIVA-ISS. We confirmed the effectiveness of the proposed method through dynamic experiments and that the BCG-simulated signal and road noise could be separated under a moderate load on the piezoelectric film sensor.