Abstract:As the number of heterogeneous redundant sensors on unmanned aerial vehicle (UAV) increases, onboard sensors require a more rational and efficient credibility evaluation system and a resilient fusion framework to achieve the essence of seamless sensor group switching. A simple and efficient sensor credibility evaluation system is proposed to guide the selection of the optimal multi-source sensor submodel combination, thereby providing key model prior knowledge for multi-source resilient fusion. Furthermore, a multi-model interactive resilient fusion framework based on RIEKF is proposed, utilizing the defined sensor credibility indexes to guide the design of the model transition probability matrix, thereby reducing the sensitivity of submodel weights to fusion stability and solving the problem of the model transition matrix lacking a basis for adjustment. Model weights are updated in real time through credibility prior information and submodel posterior probabilities, thus leveraging the adaptive resilience advantage between models to achieve seamless switching between submodels in complex environments. Experimental results show that the algorithm presented in this paper, without using any sensor fault diagnosis and isolation logic, without setting any complex detection timing and thresholds, demonstrates a resilience advantage, thereby enhancing the adaptability of the state estimation system in complex environments.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to establish a more reasonable and efficient sensor credibility assessment system and a resilient fusion framework on unmanned aerial vehicles (UAVs) with the increase in the number of heterogeneous redundant sensors, in order to achieve seamless sensor group switching. Specifically, the paper mainly focuses on the following issues: 1. **Sensor Credibility Assessment**: Traditional sensor fusion frameworks usually focus on a single combination and ignore the coordination between sensors. The strong heterogeneity in function and error characteristics among different heterogeneous sensors poses significant challenges to the design of fusion algorithms based on traditional methods. Therefore, a simple and effective sensor credibility assessment system is required to guide the selection of multi - source sensor sub - model combinations. 2. **Resilient Fusion Framework**: Existing fusion algorithms usually require steps such as validity detection, fault diagnosis, diagnosis and isolation, which increase the difficulty of algorithm design and debugging. In addition, threshold - dependent methods are prone to false positives and missed detections, and this method becomes extremely complex as the number of heterogeneous sensors increases. Therefore, a multi - source resilient fusion mechanism is needed that can improve the adaptability and resilience of the state estimation system without using complex fault diagnosis logic. 3. **Design of Model Transition Probability Matrix**: Although the interacting multiple - model (IMM) method can "soft - switch" different models, its fusion accuracy is very sensitive to the model transition probability (i.e., π in the Markov chain). In current practice, these transition probabilities are usually adjusted manually, and inappropriate transition probabilities will significantly affect the fusion results. Therefore, it is necessary to guide the design of the filter transition probability matrix according to the current credibility of heterogeneous sensors to reduce the influence of sub - model weights on fusion stability. ### Main Contributions of the Paper 1. **Proposed a simple dimensionless sensor credibility assessment system**: Select the optimal and sub - optimal multi - source sensor sub - model combination architectures according to the credibility of different heterogeneous sensors, providing model prior information for multi - source resilient fusion. 2. **Proposed an improved multi - source resilient fusion algorithm**: Use the invariant Kalman filter as a sub - model and design a multi - model transition probability matrix through the sensor credibility index, thereby reducing the sensitivity of sub - model weights to fusion stability. 3. **Verified the robustness and resilience advantages of the algorithm through actual flight data**: Experiments were carried out on the UAV redundant IMU/GPS platform, and multiple faults (such as heavy - tailed noise and communication rejection) were injected to comprehensively test the resilience advantages of the algorithm. ### Summary The paper aims to solve the problem of efficient fusion of multi - source heterogeneous sensors in complex environments by introducing a new sensor credibility assessment system and a resilient fusion framework, thereby improving the reliability and adaptability of the navigation system.

Multi-Source Interactive Resilient Fusion Algorithm Based on RIEKF

Aeromarine Reentry Vehicle Tracking Based on Asynchronous Data Fusion of Multi-Airborne-platform Sensors

Multi-Source Adaptive Selection and Fusion for Pedestrian Dead Reckoning

Radar and Camera Fusion for Multi-Task Sensing in Autonomous Driving

Fata Fusion Algorithm for Multi-Sensor Dynamic System Based on Interacting Multiple Model

An Interactive Multi Model prediction-based Radar-Camera Fusion Algorithm for Autonomous Vehicles

Resilient Unscented Kalman Filtering Fusion With Dynamic Event-Triggered Scheme: Applications to Multiple Unmanned Aerial Vehicles

Multi-sensor Hybrid Fusion Algorithm Based on Adaptive Square-Root Cubature Kalman Filter

The Research on Multi-source Uncertain Data Fusion

Compatible multi-model output fusion for complex systems modeling via set operation-based focus data identification

Remaining Useful Life Estimation Based on Asynchronous Multisource Monitoring Information Fusion.

Robust Fusion Estimation Under Data-Driven Transmission Strategy for Multisensor Systems With Random Packet Drops

Research on NFE model of multi-sensor information fusion

Multi-source Information Data Fusion Method under Complex Battlefield Situation

Research and Application on NFE Model of Multi-Sensor Information Fusion

Semi-Aerodynamic Model Aided Invariant Kalman Filtering for UAV Full-State Estimation

GNSS-INS-Dynamic Fusion with Robustness to Outliers Based on External Force State Estimation

To Fuse or Not to Fuse: Measuring Consistency in Multi-Sensor Fusion for Aerial Robots

Observability Analysis Based on QR Decomposition and Adaptive Information Fusion Mechanism for Low-Cost Multi-Rotor UAVs

Theoretical and simulation research on optimizing information fusion method for multi-sensor

Multi-sensor Data Fusion Algorithm Based on Fuzzy Adaptive Kalman Filter