CARSLIDE: Mapping and monitoring system for lanslides forecast

ABSTRACT

The present project aims at producing a diagnostic system called CAR-SLIDE that by integrating in situ data detected from on board sophisticated innovative measuring systems, with Earth Observation techniques (based on weather forecast, SAR and optical data) will be capable to foresee and monitor landslide events along rail networks. Particular importance will be attached to the use of advanced SAR interferometry techniques capable to detect small ground surface deformations  that in several cases precede the critical failure phase leading to landslide movements. Special attention will be directed to the use of SAR images detected by COSMO/SkyMed (ASI) constellation capable to achieve very high spatial resolutions and very short revisit and response time.

PRINCIPAL INVESTIGATOR
ETTORE STELLA

   TYPE AND DATE
   NATIONAL PROJECT
   STARTING DATE – Jul 1, 2011
   ENDING DATE – Nov 30, 2014

COST
TOTAL COST €: 5.730.000,00
TOTAL FUNDING €

CNR ISSIA
COST € 630.000,00
FUNDING € 540.000,00

SUMMARY

For all the European Countries, the rail network represents a key critical infrastructure, deserving protection in view of its continuous structure spread over the whole territory, of the high number of European citizens using it for personal and professional reasons, and of the large volume of freight moving through it.
Railway system, in general and the Italian rail system in particular,  traverses a wide variety of terrains and encounters a range of geo-technical conditions. The interaction of these factors together with climatic, and seismic forcing, may produce some geo-technical failures that impact on the safety and efficiency of rail operations.
In such context, a particular interest is directed to the development of technologies regarding both the prevention of mishaps of infrastructures, due to natural disasters and/or to terrorist attacks, and the fast recovery of their normal working conditions after the occurrence of accidents (disaster managing). Both those issues are of strategic interest for EU Countries, and in particular for Italy, since, unlike other countries, it is characterized by a highly particular geo-morphological and hydro geological structure that increases the risk of natural catastrophes due to landslides, overflowings and floods.
The present project aims at producing a diagnostic system called CAR-SLIDE that by integrating in situ data detected from on board sophisticated innovative measuring systems, with Earth Observation techniques (based on weather forecast, SAR and optical data) will be capable to foresee and monitor landslide events along rail networks. Particular importance will be attached to the use of advanced SAR interferometry techniques capable to detect small ground surface deformations  that in several cases precede the critical failure phase leading to landslide movements. Special attention will be directed to the use of SAR images detected by COSMO/SkyMed (ASI) constellation capable to achieve very high spatial resolutions and very short revisit and response time.
The innovative character of the proposal is mainly the use of measuring systems detecting values that could be directly exploited as inputs for data processing systems, addressed to the identification of hazard level in an unstable slope.
The project will exploit for the first time the dynamic integration of different leading edge satellite technologies in operational context, aiming at guarantee the safety of train and railway in case of accident. The environmental information will be provided by a Space Segment (e.g. Earth Observation satellites, GPS, EGNOS and/or Galileo satellites), and rail network infrastructure information will be provided by an on board system, and integrated into a Decision Support System made of a data processor, multilevel dynamic GIS (Geographic Information System) system, a web portal and a communication layer.
The unique capability of the CAR-SLIDE vehicle to seamlessly and continuously integrate information coming from different sources will enable Railway Operators to monitor the rail infrastructure and its surroundings, during the normal course of the daily rail transport and to promptly react when an unexpected situations occurs setting an effective disaster response with thorough knowledge of the area where it has taken place.
The feasibility of the technical solution, resulting from the research activities, will be pointed out with the aid of a full-scale Demonstrator that will undergo on-site tests on a few railway lines.

Range Images for 3D Reconstruction

The analysis of indoor environments returned by range sensors for the realization of 3D maps is a topic of interest for ISSIA-CNR. Starting from robotics, 3D mapping has founds its applicability in many other fields ranging from surface quality control, micro- and macro-profiling and bin picking in mechanical industries, to medicine, geology, archaeology and reverse engineering. In that context different acquisitions of the same surroundings are investigated in order to seek for contingent spatial changes of the scene topology.
This goal is achieved by the use of point clouds. These data are commonly affected by three error sources that should be compensated: statistical white noise, colored noise due to the temperature control, and failed measurements. At the same time consecutive acquisitions may differ in the trajectory followed by the sensor: measurement may start at different initial points and straight trajectory may be tilted by an angle α.
All these problems make necessary the use of proper techniques to lighten point cloud datasets and to register them. Well-established techniques, such as the Split-and-Merge algorithm and spike representations of surface topology are proved to simplify the point clouds. Furthermore an improved iterative closest point (ICP) algorithm is developed, getting the optimization of a point-by-point distance function computed between subsampled versions of the initial point clouds. As a consequence the value reached by this function at the convergence is itself expression of how much the scene has changed between the two acquisitions under registration. Moreover, the orthographic nature of the acquisition can lead to occlusions that remove useful information in the reconstruction of surfaces. In this case implicit differences can raise even if the scene is not altered by any modification. These problems are overcome by the use of deletion masks, which can be defined for each couple of acquisitions, knowing the result of the scene registration process.
Finally the derivation of easily accessible map of the environment is accomplished through difference graphical plots. Reconstructed surfaces are resampled producing a virtual measurement of the reconstructed environment, avoiding errors and data losses, typical of real measurements. Then the outcome ranges are compared through a distance operator. In other words, each elementary surface of the initial reconstructed space is labeled with the result of the comparison returned by the operator applied to the corresponding points. Fig. 1 clearly demonstrates that, although dataset are affected by consistent errors, both intrinsic (noise, failed acquisitions, etc.) and extrinsic (different system of coordinates for each probe), only those regions that are actually involved in the plane displacement are successfully highlighted.

http://www.issia.cnr.it/wp/wp-content/themes/salient-2/css/fonts/svg/5.svg

SEE ALL SENSING, COMMUNICATION AND COMPUTATIONAL INTELLIGENCE PROJECTS

SOURCE OF FOUNDING

logo1

PARTNERSHIP

TECNOGAMMA S.p.A., Italy
VVN S.r.l., Italy
PLANETEK Italia S.r.l., Italy
GAP S.r.l., Italy
SRB Costruzioni S.r.l., Italy
CNR ISSIA, Italy
Politecnico di Bari – Dipartimento Interateneo di Fisica, Italy
Consorzio Nazionale di Ricerca per le Tecnologie Optoelettroniche Dell’InP “OPTEL InP”, Italy