Traditional WiFi fingerprint positioning normally uses a time consuming and labor intensive process called site survey. This paper proposes a crowdsourcing and multi-source fusion based fingerprint sensing (CMFS) to replace the traditional site survey approach. In CMFS, motion sensors are used to construct the radio map by volunteers and Pedestrian Dead Reckoning (PDR) based on motion sensors equipped in smartphones is used for positioning. In positioning phase, an Extended Kalman Filter (EKF) based multi-source fusion algorithm is further developed to tackle the nonlinear fusion so that both positioning accuracy and robustness can be improved. Both experimental and simulation results verify that the performance of proposed schemes is comparable to the traditional fingerprint approaches. Further the EKF based fusion scheme is found to improve smoothness and stability of CMFS greatly.

A multi-source fusion positioning system (MFPS) adopts a data fusion technique for hybrid positioning. This technique fully utilizes several types of positioning sources in contrast to the conventional positioning methods, which use a single positioning source. Sensor data collection and processing are essential components in MFPSs. A framework for a practical center-server-based MFPS is proposed in this paper. A big sensor data preprocessing (BSDP) scheme, which is composed of data extraction (DE-BSDP), data gathering (DG-BSDP), and data transmission (DT-BSDP), is further proposed under this framework. DE-BSDP attempts to remove useless positioning data from fusion sources. The big sensor data are further compressed in DG-BSDP, in which a compressive sensing technique is adopted to realize data compression before data transmission. After the data gathering phase, the compressed data are transmitted to the fusion center for reconstruction in DT-BSDP. The proposed BSDP method can reduce the data collection amount significantly and improves the data transmission efficiency with a slight reduction on positioning accuracy. Experiments and simulations verify the effectiveness of the proposed BSDP scheme.

Autonomous Underwater Vehicle (AUV) has been widely used in underwater missions. Cooperative localization (CL) is a key technology especially for multi-AUVs collaborative operations. With great demands for accurate and real-time localization, error dispersion in nonlinear fusion and information transmission difficulties caused by underwater environment limitations become challenges in multi-AUVs cooperative localization. In this paper, a federated learning (FL) framework for multi-AUVs CL is designed, based on which a novel CL algorithm combining FL and extended Kalman filter (EKF) is proposed. The proposed FL-EKF algorithm can fuse the advantages of EKF and FL adequately to realize high-precision real-time underwater cooperative localization in long-duration operations. Simulations and experiments are conducted to verify the performance of the proposed algorithm.

Underwater high precision positioning and navigation is a key technical support for underwater accurate operation of Autonomous Underwater Vehicle (AUV). Traditional AUV localization adopts Acoustic Positioning System (APS) mostly, and Long Baseline (LBL) is one of the most widely used. LBL can achieve a high positioning accuracy, but multiple acoustic beacons need to be deployed. The deployment, calibrating, and recycling of those beacons are time-consuming and labor-intensive. This paper focuses on Range-Only Single-Beacon (ROSB) technology to realize AUV localization. Compared to LBL, ROSB can reduce the time and labor of deploying beacons greatly with only one beacon adopted. Aiming at the weakness of initial observation capacity in ROSB, a virtual beacon and adjustment theory based initial location calculating algorithm is proposed. Furthermore, a ROSB based Underwater Multi-sensor Fusion Positioning (UMFP) method using factor graph is presented, in which ROSB is fused with other sensors equipped on AUV. Experiments and simulations show that the ROSB based UMFP scheme can not only save time and labor greatly compared to LBL, but also achieve a high positioning accuracy.

Singe-beacon localization technology can help Autonomous Underwater Vehicles (AUVs) to obtain precise positions by deploying only one beacon. It is considered as a promising way, benefiting from saving much time and labor compared with traditional Long-Baseline Localization (LBL). A typical single-beacon localization scheme contains two essential questions: the initial observability problem and long-endurance trajectory tracking problem. Aiming at these core problems, a comprehensive solution for single-beacon localization is described in this paper. An multi-hypothesis initial position discriminant method is proposed firstly, which helps to achieve accurate initial location based on observability analysis. Then, an Adaptive Network Fuzzy Inference System (ANFIS)-improved Extended Kalman Filter (EKF) method is proposed, in which single-beacon measuring information is fused with off-the-shelf sensors, including DVL, Compass, etc. ANFIS-EKF can help to improve trajectory tracking precisions by restraining the heavy loss of linearization in conventional EKF. Both simulation and field tests are conducted to verify the performance of the proposed algorithms.