Trust-Region Neural Moving Horizon Estimation for Robots

TL;DR

This paper introduces a trust-region training method for neural moving horizon estimation (NeuroMHE) in robots, significantly improving disturbance estimation accuracy and efficiency with minimal data and computational resources.

Contribution

It develops a second-order derivative computation (MHE Hessian) for NeuroMHE, enabling trust-region optimization that enhances training speed and robustness over traditional gradient descent methods.

Findings

Achieves training in under 5 minutes with only 100 data points.

Outperforms state-of-the-art neural estimators by up to 68.1% in force estimation accuracy.

Uses only 1.4% of the parameters of previous models.

Abstract

Accurate disturbance estimation is essential for safe robot operations. The recently proposed neural moving horizon estimation (NeuroMHE), which uses a portable neural network to model the MHE's weightings, has shown promise in further pushing the accuracy and efficiency boundary. Currently, NeuroMHE is trained through gradient descent, with its gradient computed recursively using a Kalman filter. This paper proposes a trust-region policy optimization method for training NeuroMHE. We achieve this by providing the second-order derivatives of MHE, referred to as the MHE Hessian. Remarkably, we show that much of computation already used to obtain the gradient, especially the Kalman filter, can be efficiently reused to compute the MHE Hessian. This offers linear computational complexity relative to the MHE horizon. As a case study, we evaluate the proposed trust region NeuroMHE on real quadrotor flight data for disturbance estimation. Our approach demonstrates highly efficient training in under 5 min using only 100 data points. It outperforms a state-of-the-art neural estimator by up to 68.1% in force estimation accuracy, utilizing only 1.4% of its network parameters. Furthermore, our method showcases enhanced robustness to network initialization compared to the gradient descent counterpart.