State estimation of multi-sensor systems based on error-state Kalman

With the rapid advancement of multi-sensor systems, the capabilities of robots in complex scenes are gradually improving. A multi-sensor system state estimation algorithm based on error-state Kalman filter is proposed to address the issues of noise interference, sensor data loss, and interference from moving targets in dynamic scenes. Firstly, the state estimation method of multi-sensor system based on sequential fusion framework is designed to effectively integrate and process different sensor data. On this basis, further research is conducted to design a lightweight detection algorithm based on multi-sensor data for identifying and processing moving targets in dynamic scenes, thereby reducing their interference with state estimation. Finally, a sequential fusion odometer based on error-state Kalman filtering is constructed to enhance the accuracy and stability of state estimation, and further optimize the performance of the entire state estimation system. Experimental results show that the proposed algorithm achieves an estimation error of only 0.36, significantly outperforming the comparison algorithms. The mean average precision on the KITTI and NuScenes datasets reaches 0.89 and 0.85, respectively. The algorithm maintains stable efficiency across varying data scales, with low packet loss rates and controllable false detection rates in medium-to-long-term state estimation. Packet loss rates in noisy environments and dynamic target interference scenarios are 0.53% and 1.07%, respectively. The multi-sensor fusion state estimation algorithm proposed in the study can effectively handle the interference problem in dynamic scenes, significantly improving the localization and mapping performance of robots in complex environments. The research provides an effective solution for the stable positioning and mapping of robots in complex dynamic environments, which is of great significance for improving the application performance of robots in fields such as autonomous driving and special operations.