#	Pagina
attuale pagina	/open-h2020/projects/199152/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 2 - AEROBI (AErial RObotic System for In-Depth Bridge Inspection by Contact)

Teaser

Summary

AEROBI project intends to develop a robotic system to inspect the bridges and assess their status. The project is based on the development of an aerial unmanned system equipped with sensors that can detect, identify and measure the anomalies and defects. The detection and identification are supported by a computer vision module analysing the images taken by cameras. The various measurements are done with lasers and an innovative Ultra-Sonic sensor. As the US sensor must be in contact with the bridge for the measurements, the UAV is equiped with an articulated robotic arm that can position the sensor accurately on the cracks. After the collection of the anomalies and defects found and measured (surface defects and internal rebars defects), the decision support sub-system will analyse the bridge status and propose adapted actions.
The goal of the system is therefore to perform as automatically as possible the bridge inspection (presently essentially manual) and to analyse the health of the bridge (presently performed by experts). The initial objective was to detect, identify and qualify the defects. The project being largely user-driven, the system capabilities were extended to the analysis of dynamic deformations of the bridge.
AEROBI is of utmost importance for the organisations in charge of the management of road/railways networks. The present cost of bridge inspections cannot allow for frequent inspections, so the bridge monitoring is a compromise between available budget and risks. The operational cost of AEROBI system is expected to be one level of magnitude lower than the manual inspection and assessment and can therefore allow for more regular and complete inspections. The bridge maintenance woukl therefore switch from a curative maintenance to a preventive maintenance.
The main objectives of the projects are:
1. Technical: Develop the unmanned platform that can integrate the adapted sensors and an articulated robotic arm to position accurately the ultra-sonic sensor specially developed for the project. Develop/adapt the Ground Control Station to pilot the platform, control the missions and exploit the data. Integrate the whole system (platform, data link and GCS) to produce an user-friendly operational system easily deployable on bridge sites.
2. Operational: Beside the manual pilotage skills needed, the system deployment and operations shall be done by civil engineers and inspection experts without deep skills in robotics and ITC domains. The system shall also be compatible with EU (and abroad) rules and constraints to operate unmanned platforms in the airspace without the need to perform long and heavy procedures to obtain flight authorisations.
3. Business: The development of AEROBI system is permanently checked against the affordability objectives. The project needs also to define clearly the best business model possible to cope with the inspection business characteristics.
The system is complex as it integrates a lot of different technologies with a high accuracy/performance dimension. The main issue is the navigation capabilities of the UAV. The flight without GNSS necessitates to combine various technologies and the positioning accuracy is thus variable. The UAV also carries a lot of different payloads. So the aerial platform needs to be large (also for outdoor airworthiness) while remaining under the 20 kg for the flight authorisations. The computer vision sub-system needs both an accurate positioning and a sufficient stability of the platform. This complexity led to an incremental approach with two operational versions V2-A and V2-B that will be tested on two operational bridges at M26 and M32. In order to mitigate the risks of delays, an additional test campaign was added at M12 in Spain. The objective of it was to verify that the combination UAV and computer vision was sound, although the system was not completely integrated. The results were more than encouraging since all anomalies (even unimportant ones)

Work performed

The work performed from the beginning of the project was in the following domain:
1. End-user requirements, system specifications, system design:
The end-user requirements (D1.1) was elaborated as a reference document. It comprises not only the user requirements for AEROBI but also all the requirements for bridge inspection. In addition, the document contains a taxonomy of all bridges types and a proposition of standardised bridge ontology/identity card. D1.2 presents the AEROBI system specifications and design.
2. UAVs and GCS: the AEROBI final UAV platform is currently close to being finished at CATEC premises. It is derived from previous prototypes, but due to its outdoor utilistion and the payload it will have to carry, it is nuch larger. Flight test will start soon. The hybrid navigation system is currently under development. The GCS already combine piloting capabilities and computer vision exploitation.
3. Computer vision: the first version of the computer vision module that was tested on an real bridge in Spain in December 2016 showed satisfactory results in terms of images and identification of cracks. The module is currently being enhanced and optimised for the classification and identification of the anomalies.
4. Sensors: The ultra-sonic sensor version 1 is completed at around 80% and tests will start soon to integrate it on the articulated arm. The total station is also progressing well.
5. Decision support module: the module is currently in the integration phase.
6. Integration: The integration platforms are available. There was already one initial system integration for the tests in December 2016.
7. Trials/test campaigns: A first trial took place south of Seville in December 2016.
8. Dissemination. The progress is good and AEROBI has already reached a large public, end-users, stakeholders and experts.
9. Exploitation: Although the outcomes of the project are still preliminary, the business case and the exploitation roadmaps are already very promising.

Final results

Sensors: the ultra-sonic sensor, following previous efforts in ROBO-SPECT project, has now reached a good maturity, using a new technological approach that is both more stable and more reliable. The tests have shown results that are comparable to manual measurements.
UAV - navigation: in AEROBI, the UAV missions are essentially performed without GNSS (bridge mask). The development of the hybrid navigation system combining inertial, laser measures and imagery is progressing well.
UAV - platforms: Due to the AEROBI specifications (sensors, articulated arm, airworthiness, reliability), the platform is specific and is being developed by CATEC and USE. The physical design is innovative and will lead to one patent.
Computer vision: the image analysis to detect anomalies and defects combines the classical method with deep learning techniques. The results of the tests performed during the preliminary trials showed excellent performances in detection.
Decision Support: the integration of all the modules that will enable a deterministic status assessment of the bridge is close to completion.

The system in its present status and with the preliminary results obtained during the prliminary trials has already given a lot of confidence to the end-users for the final outcomes.