An Innovative Hard-Time Synchronization System for Multisensor Measurement Applications

Real-time and precise time synchronization are essential for multi-sensor fusion in simultaneous localization and mapping (SLAM) systems. Despite advances in multi-sensor time synchronization, achieving accurate, stable, and real-time synchronization is still difficult due to uncertainties like independent sensor internal clocks, transmission delays, and computational latencies. To address this issue, the authors propose a multi-sensor hard-time synchronization system (MS-HTSS) that provides deterministic time synchronization for SLAM multi-sensor fusion. A global navigation satellite system disciplined oscillator (GNSSDO) and quartz oscillator provide a unified time reference, and a ZYNQ-based hard synchronization adapter is used to synchronize high-frequency low-data-volume sensors (e.g., inertial measurement units (IMUs)) and low-frequency high-data-volume sensors (e.g., cameras). Compared to existing IMU data acquisition methods and multiple SLAM datasets, MS-HTSS’s IMU data has an average time interval of 4 ms, a time interval standard deviation of 0.001 ms, and a relative standard deviation (RSD) of 0.024%, demonstrating superior temporal precision, real-time performance, and robustness. The adapter performs hardware-level timestamping by triggering cameras with IMU signals and calculating precise image capture times, thereby mitigating transmission- and OS-induced timing errors. Experimental evaluations of FAST-LIO2 and VINS under the configurations adopted in this study demonstrate that temporally synchronized sensor data yield better trajectory accuracy and stability than unsynchronized ROS2 data. These results verify that the proposed MS-HTSS can improve the performance of open-source tightly coupled SLAM algorithms. The relevant design documents, source code, and test datasets are publicly available at https://github.com/EstherqqLiu/MS-HTSS.git.