A Method based on the Kalman Filter to Solve the Multi-sensor Data Loss Problem

Multi-sensor systems are widely deployed in autonomous driving, industrial control, and environmental monitoring, yet data loss due to sensor failures remains a critical challenge. To address this issue, this paper proposes a novel Kalman filter-based framework for multi-sensor data loss recovery. In this framework, the system model is dynamically integrated with the observed data from related sensors. The missing information is then reconstructed through the recursive calculation and data fusion capabilities of the extended Kalman filter. Specifically, the paper derive a minimum variance coefficient to optimize fusion synchronous sensor measurements such as radar position and velocity data through coordinate transformations and the inner relation of the measured object. The paper apply the theoretical analysis to derive the coefficients that minimize the estimated variance under known conditions, which show that this method can effectively reduce the estimation error caused by data loss while maintaining the computational efficiency. Compared to traditional interpolation or neural network-based approaches, the framework provides a lightweight and precise solution for real-time applications in resource-constrained environments.