Scene Interpretation and Behavior Planning for Driver Assistance U. Handmann, I. Leefken and W. v.Seelen Institut f¨ ur Neuroinformatik Lehrstuhl f¨ ur theoretische Biologie Ruhr-Universit¨atBochum 44780 Bochum, Germany ABSTRACT The scene interpretation and the behavior planning of a vehicle in real world traffic is a difficult problem to be solved. If different hierarchies of tasks and purposes are built to structure the behavior of a driver, complex systems can be designed. But finally behavior planning in vehicles can only influence the controlled variables: steering angle and velocity. In this paper a scene interpretation and a behavior planning for a driver assistance system aiming on cruise control is proposed. In this system the controlled variables are determined by an evaluation of the dynamics of a two-dimensional neural field for scene interpretation and two one-dimensional neural fields controlling steering angle and velocity. The stimuli of the fields are determined according to the sensor information. Keywords: Driver Assistance, Intelligent Vehicles, Neural Field Dynamics 1. INTRODUCTION Driver assistance systems have to assist the driver of a vehicle in his actions. The degree of the complexity of such systems can vary from information proposed to the driver 1 to almost autonomous driving. 2–4 For the given purpose the generated behavior-advice or action is determined by the actual task, the safety- and the comfort-considerations. Those constraints combined with the information about the environment build the basis of the scene interpretation and the behavior planning part of a driver assistance system. The information about the environment is gained from sensor data, knowledge and integration over time. In the scene interpretation task the incoming data (in sensor coordinates) are transformed to a common description base. Actually object- and lane-information are transformed to world coordinates with respect to the moving observer. The positions of the detected objects are determined in a bird’s-eye view of the driving plane. This representation is organized dynamically using a neural field 5,6 for stabilization of data and interpretation of the constellation of the given objects for the scene analysis. The behavior planning task has to be made up of a set of basic behaviors (e.g. tracking of a leader, driving backwards) or, if no adequate basic behavior is known in advance, by calculating a dynamic transition of the controlled values for the behavior planning. In the presented paper the dynamics for behavior planning are formulated in the coordinates of the controlled values of the vehicle. The steering angle and the velocity are controlled using two one-dimensional neural fields. 7 In the presented paper first a short introduction to the architecture in which the scene interpretation and the behavior planning are embedded is given. Then the theoretical approach to neural field dynamics and the properties of the dynamics in the one- and two-dimensional case are provided. Afterwards the structure of the scene interpretation and the considerations of the behavior planning are introduced. The paper is completed by the presentation of results based on real data as well as on simulation results and by conclusions. Correspondence: I. Leefken: E-mail: Iris.Leefken@neuroinformatik.ruhr-uni-bochum.de U. Handmann: E-mail: Uwe.Handmann@neuroinformatik.ruhr-uni-bochum.de W. v.Seelen: E-mail: institut@neuroinformatik.ruhr-uni-bochum.de