Autonomous utility service vehicles (AUSVs) represent a distinct class of automated platforms designed to operate at very low speeds (often around 5–10 km/h) for urban road maintenance tasks, yet their integration in mixed traffic remains poorly understood. This study evaluates the effect of AUSVs through a microscopic simulation using a real-world network modeled on the Pangyo autonomous driving testbed in South Korea. Twelve scenarios (six AUSV speeds, 5–30 km/h, and two traffic volumes, level of service LOS C and LOS D) were simulated; LOS C and D were deliberately selected—beyond the typical off-peak LOS A/B context—to empirically identify the demand threshold above which AUSV effects become operationally significant. Performance was assessed across vehicle, lane, link, and network levels using speed, shockwave propagation, time-to-collision (TTC) ratios, and lane-change rates. Results revealed consistent patterns. At the lane level, AUSVs at 5 km/h induced backward-moving queues up to 120 m in length and critical TTC ratios of up to 0.50 under LOS D, while no significant disturbances occurred under LOS C—revealing a critical operational threshold between the two demand regimes. At the link level, speed reductions of ≥20 km/h concentrated on two-lane bottlenecks, where overtaking opportunities were constrained. Network-wide, average speeds increased monotonically with AUSV speed, while lane-change rates peaked at 5 km/h (32.5 events/km under LOS C; 59.1 events/km under LOS D) and declined thereafter. These findings support demand-aware scheduling (off-peak for ≤10 km/h operation), minimum speed thresholds (≈15 km/h), and bottleneck-avoiding routes as practical strategies for sustainable AUSV deployment.
Kim et al. (Fri,) studied this question.