Whereas driver assistance systems were previously intended for comfort or safety, the boundaries are rather more fluid these days. 'Intelligent Drive' links sensors and systems to add a new dimension to motoring. The new S-Class is equipped with a multitude of new systems to make driving even safer and more comfortable, and its intelligent systems represent the first step on the way to autonomous driving. As part of this journey, early and comprehensive validation of all relevant assistance systems in the vehicle is of ever greater importance. Within the data logging configuration, both continuous measurements and trigger measurements are implemented. A continuous measurement covers the entire duration of the journey and typically contains several hundred of the most significant measurement signals, such as data pertaining to driving dynamics such as speed, kilometer reading and geographic position. This enables the analysis of geographic distribution, exposure, driving profiles and the availability and use of certain system functions. By contrast, trigger measurements are time-limited measurements initiated by a preprogrammed condition. These trigger conditions are defined in advance on the basis of certain signal values, signal sequences and logical conditions and programmed in the data logger. Examples include assistance system interventions, system faults, potentially safety-critical driving situations and assessments by the driver. As soon as one of the trigger conditions is fulfilled, a measurement is initiated. This usually involves a circular buffer with a lead time of 20 to 40 seconds and a run-on time of 10 to 20 seconds so that data is available for before and after the time of interest. The measurement can include tens of thousands of measurement signals, video images and raw radar and camera data. Approximately 80 trigger conditions were required to validate the new functions for the S-Class. Once the vehicles have been assembled and tested, they begin the actual field testing process. Part of the fleet is operated in pilot customer testing. Another part of the fleet begins endurance tests. Here, the vehicles are driven by professional drivers on route profiles defined for the specific system. Additional test drives are performed on defined reference routes for direct comparison of different hardware and software versions. Data is collected in the EU, the USA and South Africa. At regular intervals, the measurement data is transferred from the vehicles and the systems and data loggers checked to ensure that they are in good working order. When the data arrives from the vehicles, it undergoes an automated conversion process. The data is stored on a central server for secure management. The measurements are then converted to other file formats, events within the measurements identified and characteristic values for the measurements and events calculated. All metadata generated for the measurement data is stored in a central database. The conversion process is managed via a special user interface. The development engineers have access to the database and the measurement data by means of a user interface developed specifically for this purpose. Using this, they can quickly and easily identify and analyze the measurements and events relevant to them and enter their evaluations in the database. This offers the option to manage multiple events within a measurement, including characteristic values, assessments and short videos. On the basis of the measurement data available, the system developers can draw statistically relevant conclusions as to system performance, optimize functional parameters and conduct follow-up simulations based on the raw sensor data in the case of questionable system events.
Mercedes-Benz intelligent drive: The next step towards autonomous driving
2013
15 Seiten, 6 Bilder, 3 Quellen
Conference paper
English
Mercedes-Benz Intelligent Drive: the next step towards autonomous driving
Automotive engineering | 2013
|The next step - neue Motorengeneration von Mercedes-Benz
Automotive engineering | 2012
|Online Contents | 2007
Next : Mercedes-Benz ; das Innovationsmagazin
TIBKAT | 1.2014-2016, Ausgabe 02 (Winter 2016)
|Automotive engineering | 2009
|