Steps towards an intelligent vision system for driver assistance in urban traffic

Abstract

Most computer vision systems for vehicle guidance developed in the past were tailored to the relatively simple highway scenario. Autonomous driving in the much more complex urban environment or driver assistance systems like intelligent Stop&Go are new challenges not only from the algorithmic but also from the system architecture point of view. objects one has to recognize in this environment are:The course of the lane even if it is not given by well painted markings and does not show clothoidal geometry Small traffic signs and traffic lights in a highly colored environment Different additional traffic participants like bicyclists or pedestrians Still obstacles that limit the available free space e.g. parking cars.This contribution describes our current work on these topics. It includes appropriate vision algorithms as well as the control of different vision modules following the ,,active vision"Although the urban scenario is highly complex, it is very attractive for driver assistance. Imagine an intelligent Stop&Go system that is able to behave like a human driver: it does not onlv keeD paradigm.

Stereo-based Free Space Computation in Complex Traffic Scenarios

“…Deterministic grids are basically bi-dimensional histograms counting 3D points. They are usually obtained by projecting the 3D view of measurements onto the road and counting the number of points within the same cell area ( [10], [4]). The noise properties of the 3D measurements are not explicitly modeled.…”

Real-Time 3-D Environment Capture Systems

“…Lane and road intersection detection at close range in software is presented. A stereo system is described for detection and tracking in urban traffic in [Franke et al, 1997]. He deploys detected information on the surroundings for route planning for autonomous vehicles.…”