Parametric identification and structure searching for underwater vehicle model using symbolic regression

Models for underwater vehicle explaining its relationship between movement and the force exerting on the robot permit a wide range of development to be used in control and navigation. Yet currently no general method arrives a better model with structure and parameters for vehicles automatically. Based on the empirical data, symbolic regression method inspired by natural selection is conducted to automatically detect realistic structure and parameters of vehicle model. The proposed method is completely general and does not assume any pre-existing models before identification, it can be applied out of the box to any given vehicle experiment data. To validate and compare our approach with parameter identification methods like Levenberg-Marquardt Algorithm and Genetic Algorithm, we systematically rediscover the laws underlying underwater vehicle models and neglected laws for reflect the environments. Predicted results for datasets show that we are able to find programs that are simple enough to lead to an actual accurate model for describing the mechanisms of the vehicle.