W. von Seelen

Walking pedestrian recognition

In recent years a lot of methods providing the ability to recognize rigid obstacles-like sedans and trucks have been developed. These methods mainly provide driving relevant information to the driver. They are able to cope reliably with scenarios on motor-ways. Nevertheless, not much attention has been given to image processing approaches to increase safety of pedestrians in traffic environments. In this paper a method for detection, tracking, and final classification of pedestrians crossing the moving observers trajectory is suggested. Herein a combination of data and model driven approaches is realized. The initial detection process is based on a texture analysis and a model-based grouping of most likely geometric features belonging to a pedestrian on intensity images. Additionally, motion patterns of limb movements are analyzed to determine initial object hypotheses. For this tracking of the quasi-rigid part of the body is performed by different trackers that have been successfully employed for tracking of sedans, trucks, motor-bikes, and pedestrians. The final classification is obtained by a temporal analysis of the walking process.

Image processing and behavior planning for intelligent vehicles

Since the potential of soft computing for driver assistance systems has been recognized, much effort has been spent in the development of appropriate techniques for robust lane detection, object classification, tracking, and representation of task relevant objects. For such systems in order to be able to perform their tasks the environment must be sensed by one or more sensors. Usually a complex processing, fusion, and interpretation of the sensor data is required and imposes a modular architecture for the overall system. In this paper, we present specific approaches considering the main components of such systems. We concentrate on image processing as the main source of relevant object information, representation and fusion of data that might arise from different sensors, and behavior planning and generation as a basis for autonomous driving. Within our system components most paradigms of soft computing are employed; in this article we focus on Kalman filtering for sensor fusion, neural field dynamics for behavior generation, and evolutionary algorithms for optimization of parts of the system.

Vision-based car-following: detection, tracking, and identification

The authors have developed a vision system for automatic car following and object classification. The CARTRACK system can reliably detect, track, and identify the back of an automobile in a dynamic image taken from a following car in the same lane or in a neighbor lane. The detection and tracking system exploits the symmetry property of the rear view of normal cars. The class of objects that are detected by CARTRACK includes normal cars of all sizes as well as trucks and conventional trailers. Parallel to the tracking process, a model-based identification module identifies the type of vehicle being followed. For objects of known size, it also facilitates distance estimation. Deformable 2D models (planar feature graphs) constructed from various visual features are used. The image features in a region of interest selected by the symmetry-based detection process are matched with the model objects by means of an elastic net technique.<<ETX>>

Complex behavior by means of dynamical systems for an anthropomorphic robot

We present an architecture to generate behavior for an anthropomorphic robot. The goal is to equip the robot with the capacity to interact with a human. Motivated by the research on biological systems, our basic assumption is that the behavior to perform determines the external and internal structure of the behaving system. We describe the anthropomorphic design of our robot and present a distributed control system that generates human-like navigation and manipulation behavior. As the mathematical framework for this purpose we have developed a control system which is entirely based on dynamical systems in the form of instantiated dynamics and neural fields. We also present a dynamic scheme for the behavioral organization based on competitive dynamics.

Visual obstacle detection for automatically guided vehicles

A stereo obstacle detection system has been developed for automatically guided vehicles that operate on flat (factory) floors. The system does not attempt to reconstruct the 3D environment visually but simply tries to detect obstacles on the floor in the vehicles path. The approach to stereo image processing uses inverse perspective mappings to facilitate matching of the binocular field of vision against the expected 3D structure of the environment. Assuming a known relative camera model, a geometrical image transformation is computed which essentially compensates the stereo disparities for the image points of the floor. After the mapping operation the images are compared and local mismatches are interpreted as possible obstacle locations. The system has been successfully tested in a factory environment. The implementation runs on standard microprocessor hardware in real time.<<ETX>>

CARTRACK: computer vision-based car following

CARTRACK is a computer vision system that can reliably detect, track, and measure vehicle rears in images from a video camera in a following car. The system exploits the symmetry property typical for the rear of most vehicles on normal roads. The authors present two novel methods for detecting mirror symmetry in images, one based directly on the intensity values and another one based on a discrete representation of local orientation. CARTRACK has been used for realtime experiments with test vehicles of Volkswagen and Daimler-Benz.<<ETX>>

Real-time vehicle tracking and classification

In this paper a feature and model based approach to real-time vehicle tracking and classification is described. We proceed in two steps: 1) we establish correspondence between model and image features by an optimization algorithm; and 2) based on this correspondence, a matching vector is derived and used as input to either a Bayes classifier, a neural net or a combination of both. The current implementation updates the model parameters (position and scale) at a rate of 8-12 frames per second.

Scene analysis and organization of behavior in driver assistance systems

To reduce the number of traffic accidents and to increase the drivers comfort, the thought of designing driver assistance systems arose in the past years. Fully or partly autonomously guided vehicles, particularly for road traffic, pose high demands on the development of reliable algorithms. Principal problems are caused by having a moving observer in predominantly natural environments. At the Institut fur Neuroinformatik methods for analyzing driving relevant scenes by computer vision are developed in cooperation with several partners from the automobile industry. We present a solution for a driver assistance system. We concentrate on the aspects of video-based scene analysis and organization of behavior.

A flexible architecture for intelligent cruise control

We present a concept of a flexible and modular architecture for intelligent cruise control (ICC). The architecture can be subdivided into three different processing steps: object-related analysis of sensor data, behavior-based scene interpretation and behavior planning. Each step works on collected sensor information as well as on a knowledge base, which can be broadened by external knowledge like GPS and street maps. An intelligent car following system is described in the paper as a spin-off for behavior planning.

VISOCAR: an autonomous industrial transport vehicle guided by visual navigation

The authors describe a vision system, called VISOCAR, for mobile robot guidance in industrial environments. VISOCAR is an optically navigating AGV (automatically guided vehicle) that, owing to its local intelligence, can travel automatically in its natural environment, does not need any dedicated floor installations, and does not impose any restrictions on the route network or factory environment. It can navigate accurately on its track and integrate multisensor information to enhance functional redundancy and system reliability. It is shown that the navigation task can be broken down into a hierarchy of goals, which can be attained by functionally independent modules. The system architecture presented reflects this fact: there is a hierarchy of navigation capabilities, rather than a hierarchy of image processing steps. The successful implementation of this concept for autonomous navigation of AGVs in environments that are well structured but not artificially reduced in visual complexity shows that this modular system architecture is well suited to the specific demands of the application in spite of very limited computational resources.<<ETX>>