Xiyu Fu

Dr. Xiyu Fu

orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
31 August 2023 in Sensors
Verified
Journal Article
Open access
PDF file
SXShuchen Xu
Shuchen Xu
YSYongrong Sun
Yongrong Sun
KZKedong Zhao
Kedong Zhao
XFXiyu Fu
Xiyu Fu
SWShuaishuai Wang
Shuaishuai Wang
Satellite signals are easily lost in urban areas, which causes difficulty in vehicles being located with high precision. Visual odometry has been increasingly applied in navigation systems to solve this problem. However, visual odometry relies on dead-reckoning technology, where a slight positioning error can accumulate over time, resulting in a catastrophic positioning error. Thus, this paper proposes a road-network-map-assisted vehicle positioning method based on the theory of pose graph optimization. This method takes the dead-reckoning result of visual odometry as the input and introduces constraints from the point-line form road network map to suppress the accumulated error and improve vehicle positioning accuracy. We design an optimization and prediction model, and the original trajectory of visual odometry is optimized to obtain the corrected trajectory by introducing constraints from map correction points. The vehicle positioning result at the next moment is predicted based on the latest output of the visual odometry and corrected trajectory. The experiments carried out on the KITTI and campus datasets demonstrate the superiority of the proposed method, which can provide stable and accurate vehicle position estimation in real-time, and has higher positioning accuracy than similar map-assisted methods.

16 June 2023 in Sensors
Verified
Journal Article
Open access
PDF file
XFXiyu Fu
Xiyu Fu
YSYongrong Sun
Yongrong Sun
LWLing Wu
Ling Wu
KWKaifeng Wang
Kaifeng Wang
KZKedong Zhao
Kedong Zhao
Cycle slip detection and repair is a prerequisite to obtain high-precision positioning based on a carrier phase. Traditional triple-frequency pseudorange and phase combination algorithm are highly sensitive to the pseudorange observation accuracy. To solve the problem, a cycle slip detection and repair algorithm based on inertial aiding for a BeiDou navigation satellite system (BDS) triple-frequency signal is proposed. To enhance the robustness, the INS-aided cycle slip detection model with double-differenced observations is derived. Then, the geometry-free phase combination is united to detect the insensitive cycle slip, and the optimal coefficient combination is selected. Furthermore, the L2-norm minimum principle is used to search and confirm the cycle slip repair value. To correct the INS error accumulated over time, the extended Kalman filter based on the BDS/INS tightly coupled system is established. The vehicular experiment is conducted to evaluate the performance of the proposed algorithm from a few aspects. The results indicate that the proposed algorithm can reliably detect and repair all cycle slips that occur in one cycle, including the small and insensitive cycle slips as well as the intensive and continuous cycle slips. Additionally, in signal-challenged environments, the cycle slips occurring 14 s after a satellite signal outage can be correctly detected and repaired.

1 January 2021 in Wireless Communications and Mobile Computing
Verified
Journal Article
Open access
PDF file
LWLing Wu
Ling Wu
YSYongrong Sun
Yongrong Sun
KZKedong Zhao
Kedong Zhao
XFXiyu Fu
Xiyu Fu
Autonomous aerial refueling (AAR) technology can increase the flight endurance of unmanned air vehicles (UAVs) effectively. Drogue detection and target tracking method are significant for probe‐drogue refueling system in the docking stage. This paper proposes a novel vision‐based multistage image processing algorithm of drogue detection and target tracking for AAR. This algorithm divides the whole task into four stages: preprocessor, recognizer, predictor, and locker (PRPL). The adaptive threshold segmentation (ATS) algorithm and support vector machine (SVM) classifier are utilized in preprocessor and recognizer for drogue detection. An improved kernelized correlation filter (IKCF) tracking algorithm and scale adaptive method by window position as well as image resolution adjusted are adopted in predictor and locker for target tracking in complex dynamic environments. Finally, the proposed PRPL multistage image processing strategy is tested using an autonomous aerial refueling testbed. The results indicate that the proposed algorithm achieves high precision, good reliability, and real‐time capability compared with conventional algorithms. The average processing time is within 11 ms in various environments, which can meet the requirement for drogue detection and tracking in AAR.


24 December 2025 in Measurement Science and Technology
Verified
Journal Article
Closed access
SXShuchen Xu
Shuchen Xu
KZKedong Zhao
Kedong Zhao
YSYongrong Sun
Yongrong Sun
XFXiyu Fu
Xiyu Fu
Global navigation satellite system (GNSS) is widely used in autonomous vehicle navigation systems. Unfortunately, satellite signals are very weak or even lost in some areas, making it impossible to provide stable and high-precision localization data continuously. In GNSS-challenged regions, localization methods based on odometry can serve as effective supplements. However, they rely on dead reckoning, and the localization error will accumulate as the distance traveled increases. Thus, this paper proposes a road-network-map-assisted vehicle localization method under multidimensional constraints. This method extracts the features of the initial trajectory output by odometry and matches them with the point-line road network map to accurately locate the initial location of the vehicle. In the continuous localization stage, we view the road network map as the global sensor, introducing multidimensional constraints to correct odometry drift error and predict vehicle locations in real time. The experiments carried out on the KITTI and campus datasets demonstrate that the proposed method can locate the vehicle’s initial location in the road network map with a 5 km range and continuously suppresses odometry drift error. Compared with similar map-assisted localization methods, this approach achieves higher localization accuracy, reducing the average localization error by 14.67%.

21 July 2025 in Drones
Verified
Journal Article
Open access
PDF file
SXShuchen Xu
Shuchen Xu
KZKedong Zhao
Kedong Zhao
YSYongrong Sun
Yongrong Sun
XFXiyu Fu
Xiyu Fu
KLKang Luo
Kang Luo
Collaboration between autonomous vehicles and drones can enhance the efficiency and connectivity of three-dimensional transportation systems. When satellite signals are unavailable, vehicles can achieve accurate localization by matching rich ground environmental data to digital maps, simultaneously providing the auxiliary localization information for drones. However, conventional digital maps suffer from high construction costs, easy misalignment, and low localization accuracy. Thus, this paper proposes a visual point cloud map (VPCM) construction and matching localization for autonomous vehicles. We fuse multi-source information from vehicle-mounted sensors and the regional road network to establish the geographically high-precision VPCM. In the absence of satellite signals, we segment the prior VPCM on the road network based on real-time localization results, which accelerates matching speed and reduces mismatch probability. Simultaneously, by continuously introducing matching constraints of real-time point cloud and prior VPCM through improved iterative closest point matching method, the proposed solution can effectively suppress the drift error of the odometry and output accurate fusion localization results based on pose graph optimization theory. The experiments carried out on the KITTI datasets demonstrate the effectiveness of the proposed method, which can autonomously construct the high-precision prior VPCM. The localization strategy achieves sub-meter accuracy and reduces the average error per frame by 25.84% compared to similar methods. Subsequently, this method’s reusability and localization robustness under light condition changes and environment changes are verified using the campus dataset. Compared to the similar camera-based method, the matching success rate increased by 21.15%, and the average localization error decreased by 62.39%.