A Sigma-Point Based Uncertainty-Aware Decision Making Framework for Advanced Air Mobility

This paper presents a sample-based modular framework for incorporating uncertainty into decision-making processes for Advanced-Air-Mobility (AAM) operations without requiring modifications to existing decision-making algorithms. The a priori information used in the decision-making process is modeled as a Gaussian distribution with known mean and variance. To propagate this uncertainty, sigma points are generated using the Unscented Transform, with each sigma point treated as a distinct input instance for the decision process. The resulting outputs are then aggregated using the inverse Unscented Transform to construct the distribution of the decision outcomes. Additionally, an information-theoretic scoring function is introduced to quantify the uncertainty in the final decision. The framework is evaluated through its application to a task allocation problem for AAM operations, where the resulting assignments are compared to those obtained using only the mean input and the ground-truth information. Results demonstrate that the proposed framework produces the same assignment as the mean-only approach, while enabling a quantification of decision uncertainty.