Data Fusion of Environment-Perception Sensors for ADAS

Abstract

More and more driver assistance systems are based on a fusion of multiple environment perception sensors. This chapter gives an overview about the objectives of sensor data fusion approaches, explains the main components involved in the perception process, and explains the special topics that need to be taken into consideration in developing a multi-sensor fusion system for driver assistance systems. Focus is put on the topics of data association, tracking, classification, and the underlying architecture. The architecture strongly influences the costs, performance, and the development process of a multi-sensor fusion system. As there are no deterministic methods that guarantee an optimal solution for developing an architecture, the chapter gives an overview of established, general architecture patterns in the field of sensor data fusion and discusses their benefits and drawbacks.

Prospective and retrospective performance assessment of Advanced Driver Assistance Systems in imminent collision scenarios: the CMI-Vr approach

“…The methodology can be used both for a priori evaluations (on ADAS under development or whose operation needs to be optimized in certain road scenarios before being placed on the market) and a posteriori (as a method for evaluating the performance of an ADAS subjected to consumer/manufacturer tests). The general objective is therefore to provide the necessary tools for the evaluation of an ADAS in all phases of its life cycle, starting from SiL and moving towards Hardware-in-the-Loop (also considering the necessary physical sensor-fusion [36]), up to the tests conducted after implementation on board the vehicle. To this end, the theoretical basis of the methodology is first illustrated to subsequently present application examples of prospective and retrospective assessment of ADAS performance, useful for highlighting the approach potential.…”