Reliable absolute localisation over kilometre-scale trajectories in global navigation satellite system-denied environments remains a fundamental challenge for small autonomous aerial vehicles. Visual–inertial odometry and simultaneous localisation and mapping provide accurate short-term motion estimates but accumulate drift, requiring periodic absolute correction from external references. Two principal strategies exist: alignment to adventitious scene landmarks derived from prior mapping, and detection of deployed engineered fiducial markers. This article develops a neutral probabilistic framework to compare these approaches. We introduce Landmark Integrity as a unified descriptor of reference persistence, visibility, and identification reliability, and show that both adventitious landmark anchoring and engineered landmark anchoring reduce to robust subset selection within a field of potentially unreliable markers. A closed-form expression is derived for terminal localisation success probability as a function of usable landmark density, integrity, and required accuracy. Using literature-informed modelling priors across representative terrains and illumination conditions, we identify regimes where adventitious landmark anchoring suffices, but also identify regimes fundamentally limited by scene structure. Metre-scale terminal accuracy in low-texture or degraded-visibility environments is unlikely to be achievable by adventitious landmark anchoring alone, whereas engineered landmark anchoring maintains high reliability within a confined terminal region. These results motivate hybrid architectures combining adventitious landmark anchoring acquisition with engineered landmark anchoring or terminal seekers for precision.
No takes yet. Share an insight, caveat, or question.
Peter J. Cumpson (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: