Externally mounted sensors on autonomous vehicles can significantly affect their aerodynamic performance, especially under crosswind conditions. To improve the crosswind aerodynamic characteristics, a multi-objective aerodynamic optimization framework was developed for the shape optimization of externally mounted sensors. The Reynolds-averaged Navier–Stokes equations, coupled with the realizable k−ε turbulence model, were employed to simulate the turbulent flow, and the numerical method was validated through wind tunnel experiments. Six deformation parameters were selected as design variables, and the free-form deformation technique was applied to achieve smooth mesh deformation. Design samples were generated using the optimal Latin hypercube sampling method to analyze the relationships between the design variables and the aerodynamic responses and to identify the dominant influencing factors. Minimizing the aerodynamic drag and lateral force under crosswind conditions was defined as the optimization objective. The Non-Dominated Sorting Genetic Algorithm II based on the Kriging surrogate model was employed to optimize the sensor shapes, yielding a Pareto-optimal front. The optimal model selected from the Pareto-optimal front was compared with the baseline model. Under crosswind conditions, the drag and lateral force coefficients are reduced by 5.22% and 4.58%, respectively, demonstrating the effectiveness of the proposed optimization framework in improving the crosswind aerodynamic performance.
Zhao et al. (Sun,) studied this question.