Design of a rollover prevention controller with differential game theory and a coevolutionary genetic algorithm

This paper presents a method for designing a rollover prevention controller with differential game theory and a coevolutionary genetic algorithm. In rollover situations, a driver's steering input tends to make vehicles roll over. On the other hand, an active anti-roll bar can prevent the rollover. This situation can be modeled as a non-cooperative zero-sum two-player differential game. If an equilibrium solution to the differential game is obtained, it is guaranteed that the active anti-roll bar can prevent the rollover, even with the worst-case steering input. To obtain an equilibrium solution, this study presents a coevolutionary genetic algorithm. To show the effectiveness of this solution, simulations are performed on a vehicle simulation package, CarSim®.