Page - 63 - in Integration of Advanced Driver Assistance Systems on Full-Vehicle Level - Parametrization of an Adaptive Cruise Control System Based on Test Drives

5.1. Path Prediction Table 5.2.: Evaluation of different path prediction algorithms nk∆t [s] Jpar [m] Jcir [m] JSTM [m] JSTS [m] 3 11.9388 11.9492 11.8913 11.8222 10 26.0606 26.1204 26.3586 25.5326 wherenk is the number of predictions done at tk, meaning the number of i. All these ek form the error vector reading e=      e1 e2 ... eN      , (5.24) whereN is the number of time steps when the prediction is done, meaning the number of tk. The vector e has a length of p= N∑ k=1 nk, (5.25) which is equivalent to the number of errors used in the investigation. The evaluation is done with the cost function, which is defined as J= 1 p N∑ k=1 nk∑ i=1 ei|k. (5.26) It can be interpreted as the mean value of the distances between the measured and predicted trajectories fornk predictions atN different time steps. The comparison is done for four path prediction algorithms. The first one uses the curvature calculated by eq. (5.1) and predictes the path using eq. (5.11). The resulting cost function is called Jpar. The second one uses the same curvature but estimates the path with eq. (5.13), leading to a cost function namedJcir. The third evaluation is done with the linear STM described in chapter 5.1.2, with its cost function JSTM. The last evaluation is based on the hypothesis described in chapter 5.1.3. Its cost function is called JSTS. The evaluation was done for two different prediction horizons. The first one was set to nk∆t= 3s, and the second one to nk∆t= 10s, with the corresponding prediction distances snk = vvxnk∆t. For low vehicle speeds, the prediction distance was limited to a minimum of smin = 10m, and for high speeds to smax = 150m. The upper limit was set with respect to the maximum detection range of the environmental- recognition sensor. The number of predictions at tkwas set to pk= 100. This results in the evaluation of about p= 28.5 106 points for both cases. Table 5.2 shows the results for the two cases. 63