The worldwide diffusion and rapid escalation of smart mobile devices e.g. smartphones or tablets has led to a significant technological progress in the mobile consumer sector in the last decade. Manufacturers continuously add new sensors (with improved scanning frequency and accuracy but a lower cost) to their newly released devices, providing opportunities for the development of new innovative applications and services, which have significantly revised the way of life for our citizens. The goal of this project is designing a reliable, high precision and real-time mobile positioning system (MPS).
The present MPS faces two immediate and fundamental challenges from the real life world. First, little is known about the background and therefore accurate system modelling is difficult or even impossible and the use of conventional filters/smoothers is challenging or even infeasible. This is particularly true with the behaviour and movement of pedestrian. Secondly, the data is very rich in the sense that it comes from ‘mobile, massive multi-source and multi-type’ (4M) sensors that can have extremely fast scanning frequency. The rich sensor data can be used to circumvent poor background knowledge and to provide a greater breadth of observation regardless of individual sensor failure. However, it is expected that rich data will also pose a great challenge for real time filtering implementation.
While classical Bayes inference provides the general/standard solution, good results require correct and accurate models and few system noises and disturbances. This project develops a new concept, “Big Sensor”, to flexibly utilize massive data collections from 4M sensors in the concerned MPS such as WIFI, Bluetooth, GPS signals as well as altitude, acceleration, and direction information based on embedded software to locate mobile devices. The “Big Sensor” paradigm has identified a research opportunity to develop new theories and algorithms which do not rely on sophisticated modelling a priori but learn all knowledge from data and therefore are more reliable and accurate and even computationally faster. As a result, new estimation paradigms have been developed, including
- clustering for filtering (C4F); see publications [J1, J2, J3] blow
- fitting for smoothing (F4S); see publications [C1, C3, U1]
- data mining for joint smoothing, tracking and forecasting (JSTF); see [J8, J9, U3]
- Distributed/Networking based solutions; see [J4, J5, J6, J7, C2, U1, U2]
Smartphone-based realistic MPS applications (apps) are developed, with the potential use of European GNSS and to be integrated into a Home Care system for the purpose of tracking the movement and well-being of elderly/disabled people.
The potential use of the developed system for Home-Care/Smart-house
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 709267.
IMAGES
Li, Tiancheng; Su, Jinya; Liu, Wei; Corchado Rodríguez, Juan M. (20172017). Approximate Gaussian conjugacy: parametric recursive filtering under nonlinearity, multimode, uncertainty, and constraint, and beyond. Frontiers of Information Technology & Electronic Engineering. In press. Springer & Zhejiang University Press.
JCR (2016): 0,622
JCR (2016): 0,622
Advanced multi-sensor systems are expected to combat the challenges that arise in object recognition and state estimation in harsh environments with poor or even no prior "model" information, while bringing new challenges mainly related to data fusion and computational burden. Unlike the prevailing Markov-Bayes framework that is the basis of a large variety of stochastic filters and the approximate, we propose a clustering-based methodology for multi-object detection and estimation (MODE), named clustering for filtering (C4F), which abandons unrealistic assumptions with respect to the objects, background and sensors. Rather, based on cluster analysis of the input multi-sensor data, the C4F approach needs no prior knowledge about the latent objects (whether quantity or dynamics), can handle time-varying uncertainties regarding background and sensors such as noises, clutter and misdetection, and does so computationally fast.
Attached (C4F-oneExample) are the source code and the database for an illustrative C4F example implemented in a MODE scenario with NO prior information. The results appeared on the following work
1. T. Li, F. De la Prieta Pintado, J. M. Corchado, J. Bajo, Multi-source Homogeneous Data Clustering for Multi-target Detection from Cluttered Background with Misdetection, Applied Soft Computing 60 (2017) 436–446 @ ScienceDirect.
2. T.Li, J.M. Corchado,S. Sun and J. Bajo, Clustering for filtering: multi-object detection and estimation using multiple/massive sensors, Information Sciences, Vol. 388–389, May 2017, Pages 172-190. @ ScienceDirect.
Downloads:
In the second, the major, part, of this project, we present a continuous time state estimation framework that unifies traditionally individual tasks of smoothing, tracking, and forecasting (STF), for a class of targets subject to smooth motion processes, e.g., the target moves with nearly constant acceleration or affected by insignificant noises. Fundamentally different from the conventional Markov transition formulation, the state process is modeled by a continuous trajectory function of time (FoT) and the STF problem is formulated as an online data fitting problem with the goal of finding the trajectory FoT that best fits the observations in a sliding time-window, conditioned on a priori model information if any. Then, the state of the target, whether the past (namely, smoothing), the current (filtering) or the near-future (forecasting), can be inferred from the FoT. Our framework releases stringent statistical modeling of the target motion in real time and is applicable to a broad range of real world targets of significance such as passenger aircraft and ships which move on schedule, (segmented) smooth paths but little statistical knowledge is given about their real time movement and even about the sensors.
In addition, the proposed STF framework inherits the advantages of data fitting for accommodating arbitrary sensor revisit time, target maneuvering and missed detection. The proposed method is compared with state of the art estimators in scenarios of either maneuvering or non-maneuvering target.
The results as presented in the paper can be reproduced by running the xxx_demo main Scripts in the following attached M codes (freebody_Constrained STF_O2and maneuvering_Joint STF_O2). The source data for the results presented in the paper are included in the files as well.
Due to the need of improving the accuracy and reducing cost in indoor location system, the research group has developed a hybrid location system with a multi-agent system that uses electromagnetic fields and different types of sensors to track elements such as people or devices.
Specifically, the proposed system has solved the problem to estimating the probabilities of belonging to the points previously scanned in the intensity maps. The intensity maps obtained indoors are created using some parameters. Each of the them contains the information of electromagnetic fields such as: Wi-Fi, GSM, GPRS, RFID, Bluetooth, ZigBee networks scanned in that moment and identified by a coordinate (x,y). The rows can contain more or less columns depending on the scanned intensities. The deployment of these applications tends to take place in indoor locations such as hospitals, manufacturing plants, large warehouses or even with complementary systems such as EGNSS (European Global Navigation Satellite Systems).
The system may display heat maps, locate in real time users and devices and set different alarms depending on their location. In addition, the system allows detecting anomalous routes by defining graphs using the information of previous routes of users and devices.
Downloads: