Identification of the friction potential for the application in an automated emergency braking system
The capabilities of Automated Emergency Braking Systems (AEB) can be significantly improved when the actual friction between tires and road is known. In this work, it is investigated whether an estimation of the friction potential based on sensor data is feasible with an accuracy sufficient for an AEB. Recurrent neural networks trained by Echo State Networks (ESNs) are used to estimate friction potential from sensor data. Measurements have been conducted on a proving ground with three different tire types, two different surfaces, different driving manoeuvres and different tire inflation pressures. Standard on-board sensors of the vehicle and advanced measurement equipment have been used to measure the vehicle reaction. Based on this work, a rough understanding is gained on how well the fri ction potential can be estimated in certain situations.