Computer Vision Algorithms for the DARPA Urban Challenge 2007

Overview

The Technische Universität Braunschweig poarticipated in the DARPA Urban Challenge 2007, its autonomous vehicle 'Caroline' was among the finalists. The Computer Graphics lab provided the real-time vision algorithms for that task.

Caroline's computer vision system consists of two separate systems. The first is a monocular color segmentation based system that classifies the ground in front of the car as drivable, undrivable or unknown. It assists in situations where the drivable terrain and the surrounding area (e.g. grass, concrete or shrubs) differ in color and it deals with man-made artifacts such as lane markings as well as bad lighting and weather conditions. The second vision system is a multi-view lane detection that identifies the different kinds of lanes described by DARPA, such as broken and continuous as well as white and yellow lane markings. Using four highresolution color cameras and state-of-the-art graphics hardware, it detects its own lane and the two adjacent lanes to the left and right with a field of view of 175 degrees at up to 35 meters. The output of the lane detection algorithm is directly processed by the artificial intelligence.

The area processing unit of Caroline - Finding the way through DARPA's Urban Challenge

This paper presents a vision-based color segmentation algorithm suitable for urban environments that separates an image into areas of drivable and non-drivable terrain. Assuming that a part of the image is known to be drivable terrain, other parts of the image are classified by comparing the Euclidean distance of each pixel's color to the mean colors of the drivable area in real-time. Moving the search area depending on each frame's result ensures temporal consistency and coherence. Furthermore, the algorithm classifies artifacts such as white and yellow lane markings and hard shadows as areas of unknown drivability. The algorithm was thoroughly tested on the autonomous vehicle 'Caroline', which was a finalist in the 2007 DARPA Urban Challenge.

Read full paper [bibtex]

A Fast and Robust Approach to Lane Marking Detection and Lane Tracking

We present a lane detection algorithm that robustly detects and tracks various lane markings in real-time. The first part is a feature detection algorithm that transforms several input images into a top view perspective and analyzes local histograms. For this part we make use of state-of-the-art graphics hardware. The second part fits a very simple and flexible lane model to these lane marking features. The algorithm was thoroughly tested on an autonomous vehicle that was one of the finalists in the 2007 DARPA Urban Challenge. In combination with other sensors, i.e. a lidar, radar and vision based obstacle detection and surface classification, the autonomous vehicle is able to drive in an urban scenario at up to 15 mp/h.

Read full paper [bibtex]

vIsage - A visualization and debugging framework for distributed system applications

We present a Visualization, Simulation, And Graphical debugging Environment (vIsage) for distributed systems. Time-varying spatial data as well as other information from different sources can be displayed and superimposed in a single view at run-time. The main contribution of our framework is that it is not just a tool for visualizing the data, but it is a graphical interface for a simulation environment. Real world data can be recorded, played back or even synthesized. This enables testing and debugging of single components of complex distributed systems. Being the missing link between development, simulation and testing, e.g., in robotics applications, it was designed to significantly increase the efficiency of the software development process.

Read full paper