Risk-Aware Fault-Tolerant Multisensor Fusion for Human–Machine Decision Support in Autonomous Navigation and Mooring in Confined Waters
Authors:
- Sergey I. KondratyevSKSergey I. Kondratyev
- Evgeniy V. KhekertEKEvgeniy V. Khekert
- Nikita V. MartyushevNikita V. Martyushev
- Boris V. MalozyomovBoris V. Malozyomov
- Vladislav V. KukartsevVKVladislav V. Kukartsev
- Valeriya V. TynchenkoVTValeriya V. Tynchenko
- Tatyana Aleksandrovna PanfilovaTPTatyana Aleksandrovna Panfilova
- Yadviga Aleksandrovna TynchenkoYTYadviga Aleksandrovna Tynchenko
- Natalia I. KozhukhovaNatalia I. Kozhukhova
Open Access
Abstract
Autonomous navigation and mooring in confined waters require a navigation solution that remains reliable when individual sensing channels are delayed, unavailable, or environmentally degraded. A risk- and integrity-aware architecture was developed for joint processing of RTK-GNSS, inertial and heading measurements, short-range radar, LiDAR, camera, AIS, ultrasonic ranging, propulsion feedback, environmental, and mooring-line tension data. Asynchronous time alignment is combined with sensor-quality assessment, innovation-based fault detection and isolation, covariance adaptation, active-set reconfiguration, protection-level monitoring, and risk-dependent allocation of authority between automation and the operator. The system was evaluated during a five-day campaign comprising 40 runs, four operational phases, 480 synchronized evaluation epochs, and 16 controlled single-sensor or combined sensor-degradation events. Under nominal conditions, horizontal-position and heading RMSE were 0.043 m and 0.176°, respectively. All 64 fault-active diagnostic records were identified; mean event-log detection and controlled-recovery latencies were 3.17 s and 6.34 s. Horizontal protection-level coverage was 99.3% for single-fault epochs and 100% for combined-fault epochs. One combined-fault docking run contained five consecutive aborted decision epochs, while no unsafe-autonomy event was recorded. The measurements therefore indicate bounded degradation of the navigation solution and conservative transfer of authority when sensing integrity decreases. Because all injected-fault runs were acquired under adverse weather whereas nominal runs were acquired under calm or moderate conditions, the condition-class RMSE differences reported below are descriptive and must not be interpreted as isolated causal effects of sensor faults.Keywords
This publication has 62 references indexed in Scilit:
- A survey on maritime perception datasets and technologies for autonomous surface vesselsIntelligent Service Robotics, 2026
- Multi-vessel target tracking with camera fusion for unmanned surface vehiclesInternational Journal of Naval Architecture and Ocean Engineering, 2024
- Stochastic modeling with robust Kalman filter for real-time kinematic GPS single-frequency positioningGPS Solutions, 2023
- Robust state and protection-level estimation within tightly coupled GNSS/INS navigation systemGPS Solutions, 2023
- Adaptive Kalman filter based on integer ambiguity validation in moving base RTKGPS Solutions, 2022
- GNSS/INS Tightly Coupled Navigation with Robust Adaptive Extended Kalman FilterInternational Journal of Automotive Technology, 2022
- A new multi-sensor fusion approach for integrated ship motion perception in inland waterwaysMeasurement, 2022
- Integration of GNSS and INS with a phased array antennaGPS Solutions, 2017
- Multisensor data fusion: A review of the state-of-the-artInformation Fusion, 2013
- Study on collision avoidance in busy waterways by using AIS dataOcean Engineering, 2010