Evaluating mobility vs. latency in unmanned ground vehicles

As the penetration levels of unmanned ground vehicles (UGVs) in military applications increase, there is a growing need to evaluate their mobility across different latencies and various modes of operation ranging from pure teleoperation to full autonomy. State-of-the-art tools to evaluate mobility of ground vehicles do not address this need due to their not accounting for UGV technologies and the associated latencies. Although the trade-off between latency and performance has been thoroughly studied in the telerobotics literature and the results may qualitatively shed light onto the UGV domain, as well, a quantitative generalization is not possible due to the differences in context. Recognizing this gap, this paper presents a functional relationship between mobility and latency in high-speed, teleoperated UGVs under the context of path following. Specifically, data from human-in-the-loop simulations performed in this paper are combined with data from prior studies to span three vehicle types, three courses, and teleoperation latencies ranging from 0 s to 1 s. This combination yields for the first time a diverse data set for the context of path following in high speed, teleoperated UGVs. Based on this data set, empirical relationships are derived to quantify the trade-off between latency versus average speed and lane keeping error. These relationships can be used to establish a benchmark to evaluate the performance of autonomy-enabled UGV systems.