Integrated processing for extended target tracking using sequential Hamiltonian Monte Carlo

Tracking in high-dimensional state-space is particularly hard due to the curse of dimensionality. A way to mitigate the curse of dimensionality is the use of Sequential Hamiltonian Monte Carlo (SHMC). In this paper, we describe the exercise of tracking a single extended target using Integrated Processing based on SHMC. By way of an example, we show that this filter estimates all state variables of a high-dimensional space within reasonable time. We discuss how well the filter estimates the results for the example in terms of tracking accuracy and comment on the stability and consistency of the filter. Moreover, we mention key challenges and pitfalls to develop such a filter.