Research

Artificial objects such as engineering constructions, e.g. bridges, tunnels, and infrastructure elements, like roads, railway lines, and natural objects, e.g. dams, slopes, forest areas, make up a significant part of our environment. The regular inspection and monitoring of these objects is essential for the safety and understanding of our environment. Today, the use of multitudes of different systems is common, from tactile to optical to embedded sensors. The summary of the research and development goals in this area appears as follows:

• Research and development of novel optical system, e.g. laser scanners,  3D cameras, for the recording and monitoring of the 3D geometry of artificial and natural objects.
• Research and development of sensor implementations on mobile platforms including referencing – orientation and positioning – for rapid, highly accurate object monitoring.
• Research and development of distributed sensor systems for the cooperative recording of 3D geometry.
• Fusion of different sensors, e.g. laser scanner, camera, ultrasound, radar, to novel multi-sensor systems.

Using mobile or stationary optical sensors makes highly accurate (3D) structural monitoring of complete objects possible. Mobile systems are realized using carrier platform UAVs (“Unmanned Aerial Vehicles”), robots or humans. The generated data serves for a change analysis (deformation analysis) over time and as a digital map for the identification of potential damage areas. Aside from the use of optical sensors to record surface characteristics of an object, the fusion with sensors that make it possible to draw conclusions about the structure of the object also plays an important role.

Today, the data analysis and interpretation in complex technical processes is often a manual process: Data are analysed by humans and interpreted accordingly. High data volume is increasingly becoming a problem that can lead to highly complex situation descriptions and dependencies. Today, not only is the 3D geometry of the surfaces of artificial and natural objects recorded, but also potential deformation variables such as temperature, precipitation, load conditions, etc. The modelling of these complex interrelationships is the object of the research and development work. In the past years, there were great changes in data acquisition: the implementation of mobile systems, which operate free of global reference system, increase. New approaches, e.g. analysis of planar changes, are being prepared and implemented. The insights and results of current sensor development are pursued. At the end, actions for the preservation and safe operation of large natural and artificial structures can be derived. The summary of the research and development goals in this area appears as follows:

• Research and development of novel processes for data analysis (e.g. evaluation of data quality) while taking into account the underlying sensor system.
• Research and development of data interpretation components based on self-learning algorithms (e.g. artificial neural networks).
• Research and development of data processing methods for the evaluation of interpretation results (as a basis for the human decision-making process).

Modern sensors and sensor concepts yield an abundance of data that, after chronological and spatial referencing, yield a comprehensive description of the object. To minimize the required effort, there needs to be a focus on critical areas. The identification and location of these “hotspots” is possible on stored knowledge about the object structure (e.g. tunnel or bridge type), the critical situations and areas identified in the past and the available measurements.

The advantage of a system chain is that the corresponding strengths and weaknesses of the individual process steps are considered into account. For example, the system specifications of the used sensors are taken into account in the interpretation process and thus a faulty interpretation can be avoided (e.g. relative accuracy of single-point measurement on laser-based sensors, temperature drift on distance sensors, etc.). The summary of the research and development goals in this area appears as follows:

• Research and development of closed system chains, starting with the selection of sensors, via the linking of different complementary system components, to data fusion.
• Research and development of integrated calibration approaches for closed system chains.
• Research and development of evaluation approaches for closed system chains.

The challenge of this new approach lies in the fluent boundaries between sensor, objective and data analysis. Using a rapid, comprehensive monitoring of the object geometry, critical areas can be pre-selected. Costly measurements can be concentrated on necessary object sections. This simplifies the recording process, speeds up the entire monitoring process and reduces the costs. The tight coupling of the sensors to each other and the close connection with data analysis generates a completely new approach and significant advantages over conventional solutions with respect to the configurability, scalability and robustness of the complete solution.