Towards Human-like Machine Perception 2.0

Machine perception is a research field that is still in its infancy and is confronted with many unsolved problems. In contrast, humans generally perceive their environment without problems. These facts were the motivation to develop a bionic model for human-like machine perception, which is based on neuroscientific and neuropsychological research findings about the structural organization and function of the perceptual system of the human brain. Having systems available that are capable of a human-like perception of their environment would allow the automation of processes for which, today, human observers and their cognitive abilities are necessary. The challenge to be faced is to merge and interpret large amounts of data coming from different sources. For this purpose, an information processing principle called neuro-symbolic information processing is introduced using neuro-symbols as basic information processing units. Neuro-symbols are connected in a modular hierarchical fashion...