A Hybrid Neural-Assisted Unscented Kalman Filter for Unmanned Ground Vehicle Navigation
Authors: Gal Versano, Itzik Klein
Organizations: Autonomous Navigation and Sensor Fusion Lab, Hatter Department of Marine Technologies, Charney School of Marine Sciences, University of Haifa, Israel.
Modern autonomous navigation for unmanned ground vehicles relies on different estimators to fuse inertial sensors and GNSS measurements. However, the constant noise covariance matrices often struggle to account for dynamic real-world conditions. In this work we propose a hybrid estimation framework that bridges classical state estimation foundations with modern deep learning approaches. Instead of altering the fundamental unscented Kalman filter equations, a dedicated deep neural network is developed to predict the process and measurement noise uncertainty directly from raw inertial and GNSS measurements. We present a sim2real approach, with training performed only on simulative data. In this manner, we offer perfect ground truth data and relieves the burden of extensive data recordings. To evaluate our proposed approach and examine its generalization capabilities, we employed a 160-minutes test set from three datasets each with different types of vehicles (off-road vehicle, passenger car, and mobile robot), inertial sensors, road surface, and environmental conditions. We demonstrate across the three datasets a position improvement of 12.7% compared to the adaptive model-based approach. Thus, offering a scalable and a more robust solution for unmanned ground vehicles navigation tasks.
Figures & tables
Fig. 1 : Our proposed ANPMN-UKF implemented on the GNSS/INS fusion scansion for unmanned ground vehicles.
Fig. 2 : Examples of different simulated trajectories in 2D-view showing rectangular, sine wave, and circle trajectories.
Fig. 3 : Our dual-usage deep neural network architecture for the estimation of the inertial sensor measurement (input/output in light blue) and position measurement (input/output in pink) standard deviations.
Learning Rate
Batch Size
Epochs
Window Size
0.001
64
200
Sec window IMU
TABLE I : Main training hyperparameters values.
Parameters
Value
αANPMN
0.5
βANPMN
0.7
TABLE II : Values of tuning parameters.
Gyroscope
Accelerometer
Type Inertial sensor
Bias [deg/h]
Noise [deg/s/ Hz ]
Bias [mg]
Noise [ μ g/ Hz ]
VN-300 (ROOAD)
5
0.035
0.04
140
Arazim EXL-300 (ours)
10
0.01
0.04
300
MTi-10 (Honk-Kong)
18
0.03
0.015
60
TABLE III : IMU sensor specifications for each of dataset
Fig. 4 : (a) ROOAD dataset [ 36 ] with the Waterhod-UGV platform [ 37 ] , (b) Our recorded dataset using the ROSbot platform [ 41 ] and Arazim’s IMU EX-300 [ 42 ] , (c) Hong-Kong dataset [ 39 ] recorded using a car.
PRMSE [m]
No.Traj
UKF
MB-AUKF
ANPN-UKF
ANPMN-UKF (ours)
1
2.99
2.83
2.62
2.48
2
3.05
2.91
2.57
2.46
3
2.80
2.77
2.71
2.51
4
2.83
2.78
2.80
2.66
5
2.84
2.80
2.81
2.67
TABLE IV : Comparison of PRMSE across different trajectories in the Arazim dataset.
Fig. 5 : (a) 3D plot of Trajectory 1 showing the GT and our estimated trajectories. (b) 2D plot of Trajectory 1 showing the GT and our estimated trajectories.
PRMSE [m]
No.Traj
UKF
MB-AUKF
ANPN-UKF
ANPMN-UKF (ours)
1
5.25
4.74
4.69
3.04
2
5.58
3.73
3.68
2.92
3
5.25
3.08
3.06
2.97
4
5.28
3.77
3.33
2.89
5
3.10
2.98
2.94
2.94
TABLE V : Comparison of PRMSE across different trajectories in the ROOAD dataset.
Fig. 6 : (a) 3D plot of Trajectory 2 showing the GT and our estimated trajectories. (b) 2D plot of Trajectory 2 showing the GT and our estimated trajectories.
PRMSE [m]
No.Traj
UKF
MB-AUKF
ANPN-UKF
ANPMN-UKF (ours)
1
5.32
4.45
3.31
3.15
2
2.99
2.98
3.23
2.98
3
3.62
3.45
3.17
3.01
4
3.25
3.06
3.12
3.16
Average
3.79
3.48
3.21
3.07
TABLE VI : Comparison of PRMSE across different trajectories in the Hong Kong dataset.
Average PRMSE over all trajectories [m]
Dataset
UKF
MB-AUKF
ANPN-UKF
ANPMN-UKF (ours)
Hong-Kong
3.79
3.48
3.21
3.07
ROOAD
4.69
3.55
3.45
2.94
Arazim
2.90
2.82
2.70
2.56
TABLE VII : Average PRMSE comparison across trajectories for the proposed and baseline methods.
ANPMN-UKF (ours) Improvement [%]
Dataset
UKF
MB-AUKF
ANPN-UKF
Hong-Kong
18.98
11.78
4.15
ROOAD
37.20
17.14
14.59
Arazim
11.92
9.25
5.35
Average
22.70
12.72
8.03
TABLE VIII : Improvement percentage comparison across different methods compare to our method ANPMN-UKF.
Fig. 7 : Run-time comparison across all datasets for UKF, MB-AUKF, ANPN-UKF, and ANPMN-UKF. The boxes indicate the average runtime, while the error bars represent the minimum and maximum values.
Accurate and robust localization remains a fundamental challenge for autonomous ground vehicles. In this work, we propose a hybrid neural inertial navigation framework that integrates a wheel-mounted inertial sensors, enforced periodic trajectories, and a simple, efficient neural network capable of regressing vehicle displacement with GNSS position updates in an error-state extended Kalman filter. The periodic trajectories increase the inertial signal-to-noise ratio, allowing the network to use only inertial readings to estimate displacement. The approach is validated through real-world experiments using multiple wheel-mounted inertial sensors. Experimental results demonstrate that the proposed method achieves a significant improvement in positioning accuracy, reducing the position root mean squared error by approximately 46 % compared to standard wheel-mounted inertial sensor fusion with GNSS updates.
Gal Versano, Itzik Klein
Autonomous Navigation and Sensor Fusion Lab, Hatter Department of Marine Technologies, Charney School of Marine Sciences, University of Haifa, Israel
This work introduces a hybrid deep learning approach integrated with an Unscented Kalman Filter (UKF) to enhance pose estimation accuracy in Visual-Inertial Odometry (VIO) for autonomous navigation. The proposed model employs a Vision Transformer (ViT) network to effectively capture temporal dependencies from inertial measurement unit (IMU) data and utilizes a Multiscale Convolutional Neural Network (MCNN) to learn optical flow-based motion cues from visual data. An adaptive sensor fusion module dynamically weights IMU and visual features by leveraging estimated uncertainty, thus improving robustness in diverse and challenging environmental conditions. Additionally, a novel uncertainty-aware loss function is proposed to explicitly incorporate prediction uncertainty into the learning process, enabling robust and accurate navigation under noisy, incomplete, or unreliable sensor inputs. Comprehensive evaluations of the KITTI dataset demonstrate that the proposed method significantly outperforms baseline approaches, achieving superior performance in terms of Absolute Trajectory Error (ATE) and Relative Pose Error (RPE). The lightweight and computationally efficient model processes data at 155 FPS on an NVIDIA A100 GPU, making it highly suitable for deployment in resource-constrained autonomous systems.
Simegnew Yihunie Alaba, Yuichi Motai
Department of Electrical and Computer Engineering, College of Engineering, Virginia Commonwealth University, Richmond, VA 23220, USA
Accurate and reliable navigation is essential for autonomous ground vehicle operations. Standard INS/GNSS fusion relies on GNSS position updates, which provide limited observability of orientation and inertial sensor error states, particularly during low-dynamic motion. In this work, we propose utilizing past GNSS measurements alongside a motion model to extract meaningful vehicle acceleration information. This acceleration measurement is then integrated into the INS/GNSS filter to improve its robustness and accuracy. The proposed approach is evaluated on two real-world unmanned ground vehicle datasets collected from different mobile platforms and inertial sensor grades. Results demonstrate consistent positioning accuracy improvements relative to the standard position-aided filter, with mean position root mean square error improvements of 11.40 % and 20.74 % on the two datasets, respectively.
Gal Versano, Itzik Klein
Autonomous Navigation and Sensor Fusion Lab, Hatter Department of Marine Technologies, Charney School of Marine Sciences, University of Haifa, Israel