Page - 120 - in Maximum Tire-Road Friction Coefficient Estimation

6 Results and conclusion The estimation result for the wheel located on the low-friction surface is shown in Fig- ure 6.13, which shows a braking manoeuvre with a maximum deceleration of≈ -3 m/s2 as shown in Figure 6.14. Although the estimates deviate from the real value within the first second, the overall estimation performance is acceptable. These deviations can be partlyexplainedbythe fact thatduringabruptbrakingmanoeuvres, the localwaterfilm under the tire does not stay constant, which leads to an inaccurate reference value of µmax. In accordance with the results presented in Section 6.3, the algorithm fails to 6.5 7 7.5 8 8.5 9 9.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Reference µmax(fixed p.), no resampling µmax(sampled p.), SD limited µmax(sampled p.), ∆µmax limitedˆˆ ˆ ˆ Time in s Figure 6.13.: Estimates of µmax using fixed particles (black dotted), variable particles re-initialised based on SD (light gray) or variable particles re-initialised based on ∆µˆmax (dark gray) versus time during a braking manoeuvre with a maximum deceleration of≈−3 m/s2. Aµ split manoeuvre is performed. The result for the rear left wheel is displayed, which was located on a low-friction surface (µmax≈ 0.3). detect highµmax with low dynamic excitation, as can be seen in Figure 6.15. The com- parison of the error values shown in Table 6.2 supports the statement that low-friction conditions can be estimated more accurately. For low-friction conditions, the MAE is below 0.2 for all methods, whereas for dry roads the MAE is about 0.65. 120