Abstract: The structure of the fusion algorithm is difficult to be configured online in existing integrated navigation systems. Facing that problem, a multi-source fusion estimation framework is proposed based on the error-state extended Kalman filter (ES-EKF) and the standard observation model library. It doesn't depend on any specific platform and can be configured online, so the integrated navigation system can dynamically change the way that information sources are fused. The output information of various navigation subsystems is classified and modeled, and a standard observation model library is established, with no dependence on any specific sensor or platform. A representation rule of algorithm structure is designed to map the algorithm structure to a fusion mode and realize the flexible selection of information sources and variables. Based on the fusion estimation framework, an inertial measurement unit (IMU) / magnetometer / encoder / camera / LiDAR integrated navigation system is designed and implemented. Finally, various tests are performed on the field dataset. The system can flexibly configure the structure of the fusion algorithm by changing the fusion mode manually-statically or autonomously-dynamically, and the positioning accuracy is better than robot-localization algorithm. The experimental results show that the framework can realize multi-source fusion estimation effectively and the fusion structure can be configured online.