Abstract
In a research project between DaimlerChrysler AG and the University of Kaiserslautern an assistance system for predictive gear scheduling is developed. Environmental data (3D-route data, curves, slope, traffic signs etc.) ahead the car are used to predict the vehicle speed and after this to chose the optimal gear of an automatic transmission. Major optimisation criteria are fuel consumption and the drivers comfort.
To make efficient simulations we modelled surrounding traffic with stochastic processes. Especially Markov-Chains are suitable to create a mathematical model of such disturbances. Each source of disturbance is represented by two states for "disturbance is present" or "disturbance is not present". The transition probability is directly dependent of the traffic density. With three categories of disturbing events we generate a traffic scenario comparable to the real road traffic. Simulation of variable traffic density is realised. The results of vehicle speed simulations are compared with the average of a set of test runs on the road, that shows good congruence.
In order to make an predictive shifting decision we developed a vehicle speed prediction. It is integrated in the simulation and will be implemented in the final vehicle set-up. This module calculates online a trajectory of the future speed from the actual position to an adjustable distance (up to 2 km). Discrepancies between vehicle speed and predicted speed cause a recalculation. It needs the 3D route data and the actual position and speed of the car. The calculated trajectories are near to reality.
Together with an vehicle model this is an development environment for the predictive gear scheduling algorithm. The scheduling algorithm and the speed prediction will result in a comfortable driving assistance system.