Development of GPS time-based reference trajectories for quality assessment of multi-sensor systems

Abstract

The development of multi-sensor systems (MSSs) goes hand in hand with assessing the quality of these complex systems. Therefore, reliable reference information of superior accuracy is essential for validation, serving as ground truth. When the assessment is based on 3D point cloud comparison, appropriate reference environments with suitable geometries are required. However, validation of an MSS can also be performed directly on its 3D position or even its 6D pose. This is particularly suitable for systems without sensors for environmental acquisition. When using kinematic measurement systems, the temporal relationship between the MSS and the reference trajectory must be considered, which can be challenging. Modern sensors of an MSS are often synchronized to GPS time. However, this global time information is not provided for high-accuracy tracking sensors such as robotic total stations (RTSs) or laser trackers (LTs). Instead, only a reference to an internal sensor time is implemented, which does not meet the highest quality requirements for reference trajectories. Depending on the motion speed, time offsets in the millisecond range can lead to significant trajectory inaccuracies. This paper presents investigations for GPS time synchronization of the polar measurement elements of both RTS and LT from Leica Geosystems. While the LT uses a precise trigger signal for the time link, the RTS requires the Measure & Stream application provided by the instrument manufacturer. The two different approaches are presented theoretically. Based on empirical long-term studies, their possibilities and limitations are critically discussed. For the high-frequency LT, reference trajectories with precise synchronization based on GPS time can be reliably realized. With an RTS, improved synchronization can be achieved by using Measure & Stream. Thus, 43 % lower tangential deviations from the reference could be achieved compared to the raw data. However, there are still inconstant time offsets of 63.9 ms on average. Depending on the speed of the target, this can lead to significant position deviations. Further investigations are required. In general, the developed approaches for the realization of GPS-synchronized reference trajectories can be used not only for the quality assessment of MSS, but also for monitoring or kinematic positioning applications.