In this thesis, the development of two self-optimizing controllers for the activeguidance of an innovative rail vehicle is presented. Using actively steerable axles,the compromise between abrasive wear and stability that is inherent to passiveguidance can be suspended. A self-optimization algorithm adjusts the controllerat runtime to the current situation such that the actuation energy and the controller performance are no longer affected by disturbances. The first controllerconcept comprises a feed forward control of model based optimized trajectoriesthat are stabilised by a proportional feedback controller. The second concept isbased on a linear state feedback controller where the feedback law is optimized atruntime. Both methods share the concept of „self-optimization by multi-objectiveoptimization“. In multi-objective optimization, all requirements are formulated asa separate criterion. Subsequently, a scalar optimization problem is derived bymathematically defining the desired quantitative relation between the separateobjectives. This approach is not limited to any one controller design method.Therefore, the contribution of this thesis is not only to the presented applicationsbut further to the general method of designing self-optimizing systems.