Abstract
<jats:p>Urban canyons create a difficult paradox for autonomous ground vehicles: a vehicle may still report a precise-looking position at the moment when the underlying measurements have become least trustworthy. Multi-constellation GNSS improves availability, and 5G new radio measurements can add terrestrial geometric diversity, yet dense urban deployments also create more fault modes, including multipath, non-line-of-sight reception, base-station clock bias, geometry dilution, and simultaneous independent faults. This technical note reframes GNSS/5G positioning as an integrity problem rather than a pure accuracy problem. Built around fast advanced receiver autonomous integrity monitoring (FARAIM) for integrated GNSS/5G positioning, it explains why protection levels, solution separation, chi-square residual tests, cellular clock-bias modeling, and 5G geometry should be connected directly to vehicle behavior. Quantitative evidence is used to show why the transition from GPS-only positioning to multi-constellation GNSS/5G can be safety-relevant: in reported AGV tests, GPS-only horizontal protection levels reached hundreds of meters, whereas multi-constellation GNSS/5G reduced average HPL to approximately 4-5 m. Beyond HPL/VPL, the note argues that a practical AGV report should also include alert limit, integrity availability, continuity, missed-detection rate, false-alarm rate, time-to-alert, computation latency, and recovery or reacquisition time. These metrics turn GNSS/5G positioning from a coordinate generator into a safety interface. The central message is simple: autonomy should trust a location only when the system can bound when that location is unsafe and can translate that bound into a timely vehicle response.</jats:p>