Sonstige
This contribution introduces a number of psychological methods of analysis that are based on the practice-oriented collection of information directly at the site of an accident and that allow for an analysis and coding of the accident causes. Investigation examples and examples of the data combinations with basic medical and technical data are outlined. Objective of the collection is the inter-disciplinary investigation of human factors in the causes of accidents ("human-factor-analysis"). The psychological data are incorporated according to an integrative model for accident causes based on empiric algorithms in the data base of the accident research, where the clustered evaluation potential of comprehensive factors of the accident development can be illustrated. The central theoretical concept for the basic model of the progress of the accident from a psychological point of view comprises psychological indicators for the evaluation of the site of the accident for the analysis of the perception conditions as well as a classification of the gleaned data into the accident progress model according to chronological and local criteria. Perception conditions, action intentions and executions as well as conditions limiting perception and actions are acquired, using a questionnaire for persons involved in an accident, and are also integrated into the data structure concerning weighted feature characteristics as well as combined with other relevant features. Suitable systematization tools for the collection and coding of psychological accident development parameters have to be provided, which require primarily a model image of the corresponding processes from the persons involved in the accident (perceptions, expectations, decisions, actions). The interactive accident model contains components of the models by KÜTING 1990, MC DONALD 1972, SURREY 1969 and RASMUSSEN 1980. Based on the inter-action of the three partial systems "person", "vehicle" and "environment", the first step is the assessment of the situation by the persons involved in the accident. This is dependent on the personal attitudes and motives, on experiences and expectations concerning the progress of the situation. Subsequently, data concerning the manner of the coping with the ambiguous state as well as with the instable state (emergency reaction immediately before the accident occurs) are collected. The factors relating to the persons involved in the accident are gathered on several levels using corresponding questionnaires. The coding of the found and collected characteristics is conducted in a multidimensional evaluation relating to the technical results of the accident reconstruction and of the psychological classification, which are subsequently integrated in coded form into the data base of the accident research. The result of this analysis is a description of the development of the accident depicted on a chronological vector from a perception and decision theoretical perspective. This is explained in detail using exemplary cases.
Road condition acquisition and assessment are the key to guarantee their permanent availability. In order to maintain a country's whole road network, millions of high-resolution images have to be analyzed annually. Currently, this requires cost and time excessive manual labor. We aim to automate this process to a high degree by applying deep neural networks. Such networks need a lot of data to be trained successfully, which are not publicly available at the moment. In this paper, we present the GAPs dataset, which is the first freely available pavement distress dataset of a size, large enough to train high-performing deep neural networks. It provides high quality images, recorded by a standardized process fulfilling German federal regulations, and detailed distress annotations. For the first time, this enables a fair comparison of research in this field. Furthermore, we present a first evaluation of the state of the art in pavement distress detection and an analysis of the effectiveness of state of the art regularization techniques on this dataset.