As spotted by @Agarciafernandez, the calculation of the cross-covariance seems to use the mean weights, rather than the covariance weights. This is in contrast to [1]-[2], where the covariance weights are used.
[1] E. A. Wan and R. Van Der Merwe, "The unscented Kalman filter for nonlinear estimation," Proceedings of the IEEE 2000 Adaptive Systems for Signal Processing, Communications, and Control Symposium (Cat. No.00EX373), Lake Louise, AB, Canada, 2000, pp. 153-158
[2] Särkkä, S. & Svensson, L. Bayesian Filtering and Smoothing Cambridge University Press, 2023
As spotted by @Agarciafernandez, the calculation of the cross-covariance seems to use the mean weights, rather than the covariance weights. This is in contrast to [1]-[2], where the covariance weights are used.
[1] E. A. Wan and R. Van Der Merwe, "The unscented Kalman filter for nonlinear estimation," Proceedings of the IEEE 2000 Adaptive Systems for Signal Processing, Communications, and Control Symposium (Cat. No.00EX373), Lake Louise, AB, Canada, 2000, pp. 153-158 [2] Särkkä, S. & Svensson, L. Bayesian Filtering and Smoothing Cambridge University Press, 2023