Reference Systems for Environmental Perception
The performance of advanced driver assistance systems (ADAS) is strongly dependant on the quality of the used environmental perception sensors and algorithms.
- PDF / 624,927 Bytes
- 17 Pages / 439.37 x 666.142 pts Page_size
- 63 Downloads / 257 Views
Reference Systems for Environmental Perception Mohamed Brahmi
9.1 Introduction and Motivation The performance of advanced driver assistance systems (ADAS) is strongly dependant on the quality of the used environmental perception sensors and algorithms. Therefore, the quality of the environmental perception needs to be assessed in order to allow a formative1 and a summative2 evaluation of the ADAS perception system. This quantitative assessment of the perception sensors and algorithms requires the use of a reference system that can provide “ground-truth” information on the environment of a vehicle with higher accuracy and reliability than the used ADAS sensors. Besides accuracy and reliability, this reference system must satisfy other requirements, which will be discussed later. Reference systems can be used at different stages during the development process of ADAS systems. Based on Maurer (2013), the systematic design of driver assistance systems can be illustrated through Fig. 9.1. 1. During the specification phase of new ADAS functions, the requirements on the environmental perception are derived from requirements on the function. At this stage, the use of a reference system can help verifying whether the intended sensor system satisfies these requirements. Hence, depending on the result of this verification, these specifications can be validated or changed without having 1
Formative evaluation aims at the assessment of a system/process during the design and development. The assessment results obtained are used as feedback to improve this system/process.
2
Summative evaluation deals with the assessment of the end performance of a system/process without feeding back the results.
M. Brahmi (B) Institute of Control Engineering, Technische Universität Braunschweig, Hans-Sommer-Str. 66, D-38106 Braunschweig, Germany e-mail: [email protected] M. Maurer and H. Winner (eds.), Automotive Systems Engineering, DOI: 10.1007/978-3-642-36455-6_9, © Springer-Verlag Berlin Heidelberg 2013
205
206
M. Brahmi
Fig. 9.1 Systematic design of driver assistance systems based on Maurer (2013)
developed the whole system yet. This helps to decrease development cost and time. In addition, the given specifications of the perception sensors can be verified. 2. For the development of perception algorithms, the detailed knowledge of the sensor behavior is crucial. Through sensor interpretation and inverse models, the detection of objects in the environment can be optimized. At a further stage, where simulation of the whole system or parts of it is needed, a forward sensor model is essential to simulate the real sensor. Combined with virtual environment simulation tools and real or prototyped Electronic Control Units (ECUs), the sensor model allows the test of a system in different forms (Hardware in the Loop, Software in the Loop, Vehicle in the Loop). The development of such simulation and interpretation models can be achieved by using reference systems that provide the “ground-truth” allowing an analysis of the sensor effect
Data Loading...
