Discovering Comfortable Driving Strategies Using Simulation-Based Multiobjective Optimization

Driving a vehicle along a route consists of control actions applied to the vehicle by taking into account the vehicle and route states. Control actions are usually selected by optimizing the traveling time and the fuel consumption. However, the resulting vehicle behavior can be uncomfortable for the driver/passengers. The comfort is measured as the change of acceleration, i.e., jerk. To obtain more comfortable driving strategies, we introduce comfort as an objective to the Multiobjective Optimization algorithm for discovering Driving Strategies (MODS), thus obtaining the Multiobjective Optimization algorithm for discovering Comfortable Driving Strategies (MOCDS). The two algorithms are compared on a real-world route. The results show that MOCDS finds more comfortable driving strategies than MODS, while not significantly deteriorating their traveling time and fuel consumption. The most significant improvement in comfort is achieved on driving strategies with low fuel consumption, which are highly uncomfortable and therefore have the most room for improvement. On the other hand, the driving strategies found by MODS with short traveling time are already comfortable and therefore cannot be additionally improved.