SMARTWALK · smart Walk with Autonomous Localisation Knowhow

smart Walk with Autonomous Localisation Knowhow

The survey and analysis of personal mobility in indoor and outdoor space is a significant driver of the innovation for sustainable and responsible mobility, which is a challenge for the 21st century. But there is no practical, affordable, reliable, accurate and ethical technological solution for mapping pedestrian traffic. The good functioning of existing solutions is strongly limited by the availability and density of dedicated telecommunication network. Not only are they mainly working outdoors but they often force the user to choose a particular device, to adopt a specific behaviour and to share private data. Consequently there was a need for innovative geolocalisation processes. The project smartWALK aimed at developing new localisation algorithms based on data recorded with already available units, following an autonomous processing strategy compliant with the “privacy by design” recommendation of the European Union. Work was dedicated to the development of a wireless time synchronized localization unit labelled ULISS for Ubiquitous Localization with Inertial Signals and Satellites. This original device embeds inertial sensors (tri-axis accelerometers and gyroscopes), tri-axis magnetometers, a barometer, a GNSS receiver/antenna, a battery and a memory card in a small and lightweight box. Its design follows the European recommendation of “privacy by design” to protect location data since the recorded signals are not relying on terrestrial telecommunication networks. Conceived during smartWALK project, ULISS has facilitated the research on personal mobility with naturalistic walking. Today it opens new research ideas toward self-contained pedestrian navigation approaches. In parallel, a similar hardware, targeting a reference trajectory (footpath) measurement unit, which called PERSY for Personal Reference System has been developed. It embeds MEMS IMUs of tactical grade performance level. Being heavier and larger, it is not suitable for a handheld solution but attached on the foot. PERSY specific post-processing algorithms have been developed. They comprise diverse hybridization schemes of data available at different and sporadic time cycles including indoor GNSS carrier phase gradients, quasi static acceleration updates and magnetic field gradient update working indoors with artificial magnetic fields and geomagnetic field. A 0.3% positioning error over the travelled distance is now achieved with this solution. This result is beyond-state-of-the-art. Research was dedicated to the adaptation of existing step length models to different body sizes and shapes. Pedestrian dead reckoning (PDR) processing strategies of inertial signals combine stride frequencies with a finite number of physiological and descriptive parameters that are calibrated with training data for each person. But even under steady conditions, several physiological conditions are impacting the walking gait and consequently these parameters. Frequent calibration is needed to tune these models, prior to relying on free inertial navigation solutions in indoor locations. Two hybridization filters have been developed for calibrating the step length model and estimating the navigation solution. They integrate either GNSS standalone positions or GNSS Doppler frequencies depending on the coupling level. Experiments showed that these parameters are also changing for the same person during its journey due to fatigue or the weight of a bag, suggesting the need for frequent calibration. The errors over the estimated travelled distance, using only the handheld inertial mobile unit, are now reduced to 7% with the loosely coupled filter (10% before) and to 2% with the tightly coupled filter (5% before). 10 seconds of satellite data were found to be sufficient to calibrate the model. This time interval being rather short, it is possible to frequently use satellite data in outdoor spaces to estimate individual calibration parameters prior to penetrating indoors. Even if the quality of low cost inertial sensors and magnetometers has strongly improved, processing noisy sensor signals combined with high hand dynamics remains challenging. The estimation of accurate attitude angles for achieving long term positioning accuracy was targeted in this first period. A new Magnetic, Acceleration fields and GYroscope Quaternion based attitude angles estimation filter was proposed and demonstrated. It benefits from a gyroscope signal model directly in the quaternion set and two new opportunistic updates: magnetic angular rate update (MARU) and acceleration gradient update (AGU). A Maximum likelihood based approach has been developed to estimate the walking direction from signals collected with the handheld device that which is pointing in another direction. First stage consists in building a statistical model of the hand’s accelerations in the local horizontal plane that are correlated with biomechanics data of individual human gait. The second stage correlates real time accelerations with the model to estimate the person’s walking direction using likelihood maximization. The walking direction error, using only the inertial solution, was found to be less than 20° after 1.5 km walking, which is a beyond-state-of-the-art result. A study of existing regulations and emerging recommendations of the EU and the different national legal frameworks about the record and processing of location data has been performed. This work was conducted in collaboration with Dr. Guilbot (legal researcher) and Dr. Dommes (cognitive sciences researcher) at IFSTTAR. Europe is pioneer in the protection of individuals from personal identification through data processing since location data is recognized as personal data. But the challenges to enforce the regulation are numerous and the recommendation of “privacy by design and default” is an interesting key to achieve a universal pedestrian navigation solution. Alternative solutions are possible but they definitively require a more interdisciplinary conception, which was targeted in the project smartWALK and will be reinforced in future projects. The project outcomes have been presented at five peer reviewed international conferences (ENC, IPIN, PLANS, BioRob) and published in four peer reviewed journals (Sensors, Micromachines, IEEE Trans. on ITS, IEEE/ASME Trans. on Mechatronics, IEEE Trans on Neural Syst. & Rehab. Eng.). The dissemination activity consisted also in two invited keynotes in international conferences (ICL-GNSS 2014, IPIN 2013), two invited tutorials (IPIN 2014, IPIN 2015). Among the outreach activities are also the organization of the “portes ouvertes” of IFSTTAR Nantes, including demonstration of pedestrian navigation tools to general public (France, Sep. 2015) and the contribution to an article on indoor navigation technologies in the French daily evening newspaper “Le Monde” written by David Larousserie and titled "Naviguer dans les murs" (9 October 2013). The Marie Curie grant enabled the research fellow to successfully reintegrate Europe. smartWALK outcomes are the foundation of future projects on personal mobility. They strengthen a new research team on personal geo-positioning methods and systems for multimodal transport at GEOLOC Laboratory in IFSTTAR and enable collaboration with international navigation groups. The research fellow is now the head of GEOLOC that comprises 3 permanent persons, 4 PhD, 2 PostDocs. The Marie Curie grant ensured the reintegration in France at IFSTTAR but also a European and international scientific recognition providing a clear career development plan to the research fellow. As a result of this recognition, the IEEE sponsored IPIN International Conference will be held in France in September 2018 organized by GEOLOC.

Data: CORDIS, © European Union