Abstract：The positioning accuracy and reliability of multi-sensor integrated navigation can be improved by making full use of prior constraint information. We extend Kalman filtering under state constraints to traditional federated filtering and propose a federated filtering algorithm under state constraints. When the sub-sensor is abnormal, we use Huber method to adjust the observed noise matrix of the sub-filter based on federated filtering under state constraints. Meanwhile, we introduce adaptive information sharing factor in the information sharing stage to dynamically adjust the fusion weight of the sub-filter and obtain a robust and adaptive federated filtering algorithm with state constraints, which further reduces the impact of inaccurate sub-filter estimates on the fusion results. The method is applied to the multi-sensor integrated navigation system of strapdown inertial navigation system, GNSS and odometer. The simulation results show that the estimation accuracy of federated filter under state constraints is better than that of traditional federated filter, and robust adaptive federated filtering can further improve the accuracy and reliability of navigation and positioning under abnormal observations.