IoT for tracking of people

IoT for tracking of people

The concept Internet of Things (IoT) references the digital interconnection of common objects with Internet. IoT provides great capacity to generate important amounts of data from the physical world, which once analysed by computer tools, can be used to improve decision making as well as the performance of daily activities, thus facilitating the lifestyle of people. There are IoT solutions for many areas, such as: industrial, domestics, transport sector, etc.

Here, a solution for tracking people in external enclosures is presented. It is based on the use of an application that allows to read and send GPS data coming from the available sensors in mobile devices (smartphones), alongside the implementation the Kalman Filter in order to filter and reduce, as far as possible, the lack of precision or errors that this type of measures entails. To make the connection between the mobile device and the filter, the IoT platform Sofia2 has been used, as well as for the storage and subsequent consumption of these.

APP trajectory

IoT for tracking of people

To test this solution, we chosed the campus of the University Carlos III of Leganés, Madrid, Spain, which is a excellent enclosure surrounded by buildings to perform tests correctly and verify if the results are appropriate to the problem to be solved.

The proposed solution counts with the implementation of two variants of the Kalman filter: a classical version and the UKF (Unscented Kalman Filter) one. The first being applicable to linear systems, and the second, allows to keep in mind the non linearity of real systems. Both have as objective to model a system corresponding with the trajectory of the pedestrians while walking, allowing to correct the inherent error of GPS data coming from the mobile device used.

In the classical version of Kalman filter, the coordinates in the planes x and y, and their corresponding velocities (vx and vy) have been taken as state variables. In the case of the UKF, the state variables selected are the coordinates x and y, and the velocity module and the orientation of the pedestrian, thus considering the non lineality of the system.

In the lower figure you can see the results obtained along the entire trajectory. It is clear how the two filters implemented to soften the path traveled, correcting the small bounces that are obtained from the measurements of the sensor.


IoT for tracking of people

As a conclusion, it is clear that both filters meet the desired functionality and they have according to results to the complexity of the problem to be solved. Even though it has to be considered that the UKF allows to take into account the non-linearity of the systems by providing the information about the orientation of the pedestrian at all times during its trajectory.

In CitiSim, the tracking will be realized in the emergencies use case, being possible the Kalman Filter implementation in this situation, although instead of receiving a signal with the GPS position through a mobile device, this will be obtained by Kinect camera inside of the buildings. CitiSim provides a platform similar to Sofia2, so the main operation of this can be extrapolated to the own architecture developed for that purpose.

Links of interest:

Kalman Filter
Uncested Kalman Filter

Author: María Victoria Ortiz

Company: Abalia