Dimensionality reduction and identification of valid parameter bounds for the efficient calibration of automated driving functions

Abstract

The industrialization of automated driving functions according to level 3 requires an efficient test and calibration concept to deal with an increased complexity, growing customer demands, and a larger vehicle fleet offered. Therefore, a method for a complexity reduction of the calibration parameter space is presented. In the two-step approach, a qualitative sensitivity analysis is used to identify valid regions in the search space and subsequently decrease dimensionality based on the parameterspecific global influences. The reduced parameter space and sensitivity information can then serve as a starting point for an efficient calibration process on the target hardware. To examine the method's potential, our approach is applied to the parameter space of an automated driving function. The results expose clear dependencies between parameters and driving scenarios and allow an exclusion of parameter space dimensions based on sensitivity values. The predefined search space can be narrowed down to valid regions using the parameter range identification approach. Finally, the findings are validated with a quantitative variance-based sensitivity analysis. The validation confirms that our method provides equivalent results with a comparably smaller number of system evaluations.

Efficient Simulation Based Calibration of Automated Driving Functions Based on Sensitivity Based Optimization

“…Moreover, benefits only occur if the complexity reduction method computes a reduced parameter space compared to initially defined bounds. However, this can be expected for most vehicle control systems as concluded in previous work [20]. Theoretically, an optimizer can be freely chosen but algorithms using a discrete instead of a continuous scale contribute most to finding scenario/parameter-set-combinations whose simulations can be avoided.…”

“…On the contrary to trajectory and cell based designs which suggest a uniform sampling over each input parameter space, the radial design provides the best performance with respect to result and efficiency as exposed by the studies of Campolongo et al [23]. Convergence analyses carried out in [20] confirm that the SA provides reproducible results for different input samples using the radial approach. Thus, the number of samples for the original EEM is r • (n p + 1).…”

“…The introduced optimization approach in this work is based on a complexity reduction for an identification of influential areas in the parameter search space. The method introduced in a previously published contribution by the authors of this paper [20] deals with the same problem as addressed in this work and shown in Figure 1: Simulation based calibration of automated driving functions, i.e. finding parameter combinations for an optimal performance in a scenario catalog.…”

“…system evaluations needs to be performed. Since experimental studies have shown that for the calibration of vehicular driving systems areas at upper and lower bounds of conservatively estimated parameter ranges tend to have smaller impact [20], the initial domain can be narrowed down based on this method. After the redefinition of sensitive ranges, samples laying inside the influential hyperbox can be further processed for a global sensitivity analysis of all parameters.…”

Overview of Sensitivity Analysis Methods Capabilities for Traction AC Machines in Electrified Vehicles

“…In fact, most of the HEV/EV powertrain models are nonlinear which raises the following question, "What would be the consequences if the SA results are incorrect?" On the other hand, global sensitivity analyses (GSA) [35][36][37][38][39][40][41][42][43][44][45], which avoids the aforementioned shortcomings, are only employed by a minority of designers. From the literature published on SA-focused studies for different HEV [46][47][48] and EV [50][51][52] powertrain topologies in both commercial [51][52][53] and passenger [54][55] vehicles, a remarkable lack of the availability of data can be seen.…”