Page - 124 - in Maximum Tire-Road Friction Coefficient Estimation

6 Results and conclusion tire wear and ageing may also be identified as effects that decrease the transmittable tire forces and thus the friction potential. However, before series application can be considered, further investigations are neces- sary. Inparticular, this includes a robustand accurate determination of the longitudinal slipsx,whichwasonlypossibleusingmeasurementsof the longitudinalvelocityvxbased onaGNSS-supported inertialmeasurementplatformandresettingeachwheel’s effective tire radius. Apart from the dynamic excitation during driving, it is the accuracy of sx that limits the proposed approach. Another promising approach that is worth testing is to estimate sx andµ max simultaneously, as discussed in Section 5.3. In addition, there is room for improvement in the determination of the wheel torqueMD,i. To guarantee ro- bustness andaccuracyundermanydriving states, additional investigationsare essential. Optimizing parameters of the particle filter (e.g. by using other resampling strate- gies or adapting the proposed re-initialisation strategies) will likely further reduce the estimation error. In addition, current plans call for the development and testing of a real-time implementation of the proposed algorithm. The results presented in this chapter have to be extended by investigations, including more measurement data for each of the driving and road conditions discussed. Although there are no theoretical objections, the proposed algorithm also has to be tested with real measurements of cornering manoeuvres, with other tire types and within a higher range of dynamic excitations. Furthermore, the estimation of a global friction potential was not demonstrated with measurement data. In addition, the algorithm is currently limited to even roads. Although it would be very easy to implement a road slope and inclination within the observer model, reliable and robust estimates are necessary. The same applies to the consideration of wind forces. Vertical excitation by the road is not intended to be considered with the simplified model. However, it is also import to recog- nise that theproposedapproachhas thepotential tocontribute toasensorfusionsystem. Extending the proposed algorithm by including the lateral tire characteristics and observing µmax based on the side slip angles αi similar to Ray, [Ray97], or based on the self-aligning torque shows promise for improving the estimation accuracy for lateral driving states. However, this was beyond the scope of this work, as it was assumed that estimating µmax based on longitudinal driving states is more difficult due to the slow vehicle’s response compared to the lateral direction. In addition, it is the more likely 124