Cascade IPG Observer for Underwater Robot State Estimation

TL;DR

The paper tackles robust underwater pose estimation under GPS denial and sensor outages by introducing a cascade IPG (C-IPG) observer that first estimates orientation and then velocity and position using IMU preintegration and fused DVL/AHRS data. It treats the system as model-free and leverages a moving-horizon, Newton-like IPG procedure to avoid explicit dynamics, validating on a public underwater dataset and a BlueROV2 platform. Compared with EKF and InEKF, C-IPG achieves higher positional accuracy and lower variance, albeit with higher per-step computation. The work offers a practical, real-time capable approach for medium-cost underwater robots and provides a basis for future observer-based SLAM and re-localization strategies.

Abstract

This paper presents a novel cascade nonlinear observer framework for inertial state estimation. It tackles the problem of intermediate state estimation when external localization is unavailable or in the event of a sensor outage. The proposed observer comprises two nonlinear observers based on a recently developed iteratively preconditioned gradient descent (IPG) algorithm. It takes the inputs via an IMU preintegration model where the first observer is a quaternion-based IPG. The output for the first observer is the input for the second observer, estimating the velocity and, consequently, the position. The proposed observer is validated on a public underwater dataset and a real-world experiment using our robot platform. The estimation is compared with an extended Kalman filter (EKF) and an invariant extended Kalman filter (InEKF). Results demonstrate that our method outperforms these methods regarding better positional accuracy and lower variance.

Figures (7)

• Figure 1: An underwater vehicle in operation and an overview of the cascade IPG (C-IPG) for pose estimation using various sensors on the ROV. An example of the estimated trajectories through implementing C-IPG and InEKF is shown in red and yellow (respectively). The image shows an experiment in the Chesapeake Bay (MD, USA), where turbid water affects visibility. Hence, accurate pose estimation by inertial sensors is critical.
• Figure 2: Framework of the proposed cascade IPG observer.
• Figure 3: Comparison of 3D trajectory estimations of the vehicle from Girona dataset.
• Figure 4: Comparison for individual states of the Girona dataset for first 100 seconds. The black line indicates the ground truth (contains occasional outliers resulting in spikes). The red and blue lines represent the states estimated by InEKF and C-IPG, respectively.
• Figure 5: Comparison for individual states of the BlueROV2 for first 100 seconds. The black line indicates the ground truth. The red and blue lines represent the states estimated by InEKF and C-IPG, respectively.