3D-Safety

In view of increasing competition, port operators are faced with the challenge of operating their ports and terminals efficiently and thus with increasing automation. The increasing degree of automation with simultaneous minimization of risks in turn places special demands on the technology used.

In order to ensure personal safety in the port area, a new class of so-called 3D multilayer scanners is to be used. In order to achieve certification status, it is not only necessary to evaluate the data sufficiently quickly, but it must also be ensured that person recognition is carried out redundantly, i.e. via several sensors and two independently operating algorithms.

The goal of the project is the development of an evaluation system, which guarantees this certified person security in the port area, based on 3D multilayer scanners. In addition to the redundantly operating algorithms, this system is to have > 128-fold increased data processing, a frame rate of 10 Hz and a signal range of 100 m.

After a preparatory data processing, the data evaluation shall be developed, tested and evaluated via 4 different methodological approaches: The application of statistical algorithms, the use of a principal component analysis, the 2D image recognition via trained neural networks or the object recognition in the 3D point cloud via a trained Al are planned. Finally, the four developed algorithms are to be linked via ensemble learning in order to increase the uncertainty of the evaluation system on the one hand and to ensure the required redundancy on the other hand.

Due to the systems and functions to be developed, the safety of the logistics of goods weighing tons in a time-critical man-machine network can be significantly improved. In the future, the developed technology can also be transferred to other applications with necessary personal safety, e.g. driverless transport systems.