Implementation of Adaptive Kernel Kalman Filter in Stone Soup

The recently proposed adaptive kernel Kalman filter (AKKF) is an efficient method for highly nonlinear and high-dimensional tracking or estimation problems. Compared to other nonlinear Kalman filters (KFs), the AKKF has significantly improved performance, reducing computational complexity and avoiding resampling. It has been applied in various tracking scenarios, such as multi-sensor fusion and multi-target tracking. By using existing Stone Soup components, along with newly established kernel-based prediction and update modules, we demonstrate that the AKKF can work in the Stone Soup platform by being applied to a bearing–only tracking (BOT) problem. We hope that the AKKF will enable more applications for tracking and estimation problems, and the development of a whole class of derived algorithms in sensor fusion systems.