ResAlignNet: A Data-Driven Approach for INS/DVL Alignment

TL;DR

ResAlignNet introduces a data-driven INS/DVL alignment method using a 1D ResNet-18 to estimate the rotational offset between IMU and DVL frames directly from onboard velocity streams, achieving rapid convergence and removing reliance on prescribed maneuvers. The approach demonstrates Sim2Real transfer, showing strong alignment accuracy with minimal data (as low as 25 s) and robustness across sensor grades, outperforming traditional SVD-based Wahba solutions, especially under tactical-grade noise. Field tests on the Snapir AUV validate real-world effectiveness, with sub-degree RMSE and maximum errors below 4°, while Sim2Real training maintains competitive performance when real data is scarce. Overall, ResAlignNet advances underwater navigation by delivering fast, autonomous, sensor-agnostic alignment suitable for immediate mission deployment in diverse operational scenarios.

Abstract

Autonomous underwater vehicles rely on precise navigation systems that combine the inertial navigation system and the Doppler velocity log for successful missions in challenging environments where satellite navigation is unavailable. The effectiveness of this integration critically depends on accurate alignment between the sensor reference frames. Standard model-based alignment methods between these sensor systems suffer from lengthy convergence times, dependence on prescribed motion patterns, and reliance on external aiding sensors, significantly limiting operational flexibility. To address these limitations, this paper presents ResAlignNet, a data-driven approach using the 1D ResNet-18 architecture that transforms the alignment problem into deep neural network optimization, operating as an in-situ solution that requires only sensors on board without external positioning aids or complex vehicle maneuvers, while achieving rapid convergence in seconds. Additionally, the approach demonstrates the learning capabilities of Sim2Real transfer, enabling training in synthetic data while deploying in operational sensor measurements. Experimental validation using the Snapir autonomous underwater vehicle demonstrates that ResAlignNet achieves alignment accuracy within 0.8° using only 25 seconds of data collection, representing a 65\% reduction in convergence time compared to standard velocity-based methods. The trajectory-independent solution eliminates motion pattern requirements and enables immediate vehicle deployment without lengthy pre-mission procedures, advancing underwater navigation capabilities through robust sensor-agnostic alignment that scales across different operational scenarios and sensor specifications.

Figures (9)

• Figure 1: ResAlignNet architecture utilizing 1D ResNet-18 structure with residual connections for alignment parameters estimation.
• Figure 2: Overview of the ResAlignNet training pipeline. The GT trajectory data including velocity vector ($\boldsymbol{v}^b_{GT}$), specific force vector ($\boldsymbol{f}^b_{GT}$), and angular velocity rate vector ($\boldsymbol{\omega}^b_{GT}$) is processed through a noising pipeline to generate simulated DVL and INS velocities, which serve as input to ResAlignNet. The network estimates the alignment angles ($\hat{\boldsymbol{\theta}}$), which are compared against the GT alignment angles ($\boldsymbol{\theta}_{GT}$) using MSE loss to train the network.
• Figure 3: Detailed noising pipeline for simulation data generation. GT velocity vector in the body frame ($v^b_{GT}$) is transformed to the DVL frame using the generated GT alignment parameters ($\mathbf{C}_d^b$) and processed through the DVL error model incorporating scale factors, biases, and zero mean white Gaussian noise. Simultaneously, the inertial readings GT are processed through the INS error model with accelerometer and gyroscope errors (biases $\boldsymbol{b}_a$, $\boldsymbol{b}_g$, and zero mean white Gaussian noise $\boldsymbol{\sigma}_a$, $\boldsymbol{\sigma}_g$) before integration through the equations of motion to produce noisy INS velocity vector estimates ($\tilde{\boldsymbol{v}}^b_{INS}$).
• Figure 4: Simulated 200-second right-turn trajectory at constant speed of 2m/s.
• Figure 5: Simulation results: RMSE alignment performance for an alignment range of 0-5 degrees comparing ResAlignNet and SVD methods for turn trajectory using tactical-grade IMU.