The Illusion of Consistency: Selection-Induced Bias in Gated Kalman Innovation Statistics

2025-12-20

AI Summary

This paper identifies a previously unrecognized statistical bias in Kalman filter tracking arising jointly from validation gating and nearest-neighbor (NN) data association. Specifically, conventional chi-square gating renders the innovation process conditional rather than unconditional, inducing a systematic mean shift and a deterministic, dimension-dependent contraction of the innovation covariance. NN association further introduces an irreducible energy attenuation. The authors derive, for the first time, closed-form expressions for the first- and second-order moments of the gated-and-associated innovation under general elliptical gating, establishing an exact statistical model. Theoretical analysis and two-dimensional numerical experiments demonstrate that this dual selection mechanism induces innovation covariance biases of 10%–30%, substantially compromising filter performance evaluation and tuning of design parameters such as gate size and process noise.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Intelligent Robots: State Estimation
• Computer Vision: Motion & Tracking

Application Category

• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Validation gating is a fundamental component of classical Kalman-based tracking systems. Only measurements whose normalized innovation squared (NIS) falls below a prescribed threshold are considered for state update. While this procedure is statistically motivated by the chi-square distribution, it implicitly replaces the unconditional innovation process with a conditionally observed one, restricted to the validation event. This paper shows that innovation statistics computed after gating converge to gate-conditioned rather than nominal quantities. Under classical linear--Gaussian assumptions, we derive exact expressions for the first- and second-order moments of the innovation conditioned on ellipsoidal gating, and show that gating induces a deterministic, dimension-dependent contraction of the innovation covariance. The analysis is extended to NN association, which is shown to act as an additional statistical selection operator. We prove that selecting the minimum-norm innovation among multiple in-gate measurements introduces an unavoidable energy contraction, implying that nominal innovation statistics cannot be preserved under nontrivial gating and association. Closed-form results in the two-dimensional case quantify the combined effects and illustrate their practical significance.

Problem

• Analyzes bias in Kalman filter innovation statistics due to gating
• Derives exact moments of innovation conditioned on ellipsoidal gating
• Proves unavoidable energy contraction from gating and nearest-neighbor association

Innovation

• Gating induces deterministic contraction of innovation covariance
• NN association acts as additional statistical selection operator
• Nominal innovation statistics cannot be preserved under gating