Q: Why do we need to take integrity seriously in the vehicle environment?
A: Since the 1980s, surveyors and geodesists have used GPS for high-accuracy positioning. We take for granted centimeter- and even millimeter-level accuracy positioning capability that is faster, more reliable, at a lower cost and with fewer constraints than ever before. However, the demand for “trustworthy positioning” dismisses such achievements, and the mantra is more “availability” and greater “integrity” to support highly automated driving. Our positioning and navigation community must rise to this challenge.
A: In autonomous vehicles, a GNSS/inertial module will be just one of several sensors used for location. The risk of contributing to accidents and serious injury will be decomposed and allocated between subsystems by the OEM or system designer. Taking aviation as a model, the allocation to GNSS may be in the form of an alarm limit of a few meters with integrity risk less than 10-6/hour. However, multipath and obstructed sky make automotive risk far more difficult than aviation. Carrier-phase techniques will come into play and new approaches to protection limit estimation will be needed.
A: Advanced sensor fusion techniques now make it possible to achieve very accurate PVT results by combining multiple dissimilar sensors. Once we rely on these capabilities for autonomous driving, the primary threat to safety will come from confluences of rare events that were not observed or foreseen during system development. Design for integrity focuses attention on the identification and mitigation of potentially hazardous anomalies before they happen, not afterward.