Horizon Estimation

Horizon estimation, a crucial aspect of state estimation, aims to accurately determine the current state of a system by incorporating past measurements within a defined time window (the horizon). Recent research emphasizes the use of moving horizon estimation (MHE) coupled with neural networks (NNs), particularly in robotics and autonomous systems, leading to the development of NeuroMHE architectures. These approaches address challenges like computational complexity through optimization techniques (e.g., trust-region methods, iterative preconditioning) and data annotation difficulties in multi-sensor systems. Improved accuracy and robustness in state estimation have significant implications for control systems, particularly in applications involving agile robots and autonomous vehicles operating in uncertain environments.