analiza aiqbench
This paper introduces SUPER, a bounded controller for visual-inertial odometry that adapts measurement information at runtime to manage uncertainty and risk. While the method is computationally efficient and includes a novel action-verification protocol to distinguish controller effects from solver variability, empirical results show no general accuracy improvement across tested datasets. The study highlights the importance of rigorous verification in nondeterministic optimization environments rather than claiming broad performance gains.
SUPER - Sensitivity-based Uncertainty-aware Performance and Risk Assessment in Visual Inertial Odometry
arXiv:2512.14189v2 Announce Type: replace Abstract: This paper presents SUPER (Sensitivity-based Uncertainty-aware PErformance and Risk assessment), a bounded controller that adapts visual measurement information in visual-inertial odometry at runtime. Building on multi-view landmark dispersion from stereo observations, SUPER projects track consistency into pixel space and combines it with the reprojection residual to gate a temporally smoothed, bounded information update. Because the controller operates inside nondeterministic optimizers, we additionally propose an action-verification protocol based on factor-level logs, inactive-instrumented controls, and same-linearization replay. We evaluate SUPER in Ceres and g2o backends on EuRoC, TUM-VI, UMA-VI, and TartanAir under controlled visual corruptions, with a frozen policy and prospective confirmation. SUPER is computationally efficient and selectively active, but does not establish a general accuracy improvement: all 13 split-level confidence intervals cross zero, and effects are heterogeneous across corruption families. The action-verification protocol reliably distinguishes controller-induced state changes from solver variability.
Abstract:This paper presents SUPER (Sensitivity-based Uncertainty-aware PErformance and Risk assessment), a bounded controller that adapts visual measurement information in visual-inertial odometry at runtime. Building on multi-view landmark dispersion from stereo observations, SUPER projects track consistency into pixel space and combines it with the reprojection residual to gate a temporally smoothed, bounded information update. Because the controller operates inside nondeterministic optimizers, we additionally propose an action-verification protocol based on factor-level logs, inactive-instrumented controls, and same-linearization replay. We evaluate SUPER in Ceres and g2o backends on EuRoC, TUM-VI, UMA-VI, and TartanAir under controlled visual corruptions, with a frozen policy and prospective confirmation. SUPER is computationally efficient and selectively active, but does not establish a general accuracy improvement: all 13 split-level confidence intervals cross zero, and effects are heterogeneous across corruption families. The action-verification protocol reliably distinguishes controller-induced state changes from solver variability.
6 pages, 4 figures, 5 tables. v2: substantially revised. The Schur-complement risk indicator of v1 is replaced by a closed-loop controller adapting visual information from multi-view track dispersion, with a new action-verification protocol. ISO risk mapping and the prediction/stop-policy results of v1 are removed. v2 claims no general accuracy improvement
Robotics (cs.RO)
arXiv:2512.14189 [cs.RO]
(or arXiv:2512.14189v2 [cs.RO] for this version)
From: Johannes A. Gaus [view email] [v1] Tue, 16 Dec 2025 08:35:47 UTC (2,010 KB) [v2] Wed, 4 Nov 2026 15:38:40 UTC (984 KB)
Text integral sub licența deschisă a publicației · 224 cuvinte