A reliability-based track fusion algorithm.

Area of Science:

• Multi-sensor data fusion
• Target tracking algorithms
• Reliability-based estimation

Background:

• Traditional track fusion algorithms in multi-sensor systems suffer from accuracy-computational cost trade-offs.
• Existing methods often treat all sensor data equally, irrespective of quality, leading to suboptimal performance.
• High fusion errors occur at system track inflection points with current approaches.

Purpose of the Study:

• To develop an improved track fusion algorithm for multi-sensor, multi-target environments.
• To address the limitations of existing fusion methods, specifically accuracy, computational cost, and error at inflection points.
• To enhance state estimation by incorporating sensor data reliability.

Main Methods:

• A novel track fusion algorithm based on reliability (TFR) was developed.
• Outliers in local tracks were eliminated to improve data quality.
• Reliability of local tracks was calculated to select high-quality data for fusion.
• State estimation was performed using selected reliable tracks.

Main Results:

• The TFR algorithm significantly reduces fusion errors at system track inflection points.
• High accuracy was achieved with a lower computational cost compared to existing methods.
• Simulations demonstrated the algorithm's effectiveness, particularly in dense sensor environments.

Conclusions:

• The proposed track fusion algorithm based on reliability (TFR) offers a superior approach for multi-sensor systems.
• TFR effectively balances accuracy and computational efficiency by leveraging sensor data quality.
• The algorithm shows significant advantages in challenging dense sensor environments.