To enhance the positioning accuracy and system reliability of the INS/GNSS/ODO integrated vehicle positioning system, an adaptive unscented Kalman filter data fusion method based on the innovation detection criterion (Innovation Detection Adaptive Unscented Kalman Filter Federated Kalman Filter, IDAUKF-FKF) is proposed. When the sensors produce abnormal measurement data due to faults or interference, containing gross errors or other disturbances, this algorithm can detect the observation information based on the innovation detection criterion, identify and eliminate the abnormal measurement data. Additionally, a noise estimator is employed to adjust the statistical characteristics of the measurement noise in the local filter subsystems online, constraining the impact of local estimation errors on the global filtering. Simulation results demonstrate that the proposed multi-sensor data fusion algorithm can accurately eliminate abnormal observations and optimize local estimates. Compared to federated Kalman filtering and UKF-FKF, the positioning errors are reduced by approximately 41% and 30%, respectively, significantly improving the positioning accuracy and anti-interference performance of the INS/GNSS/ODO integrated vehicle positioning system.