Enhancement of the tipover stability of mobile manipulators with non-holonomic constraints using an adaptive neuro-fuzzy-based controller

The stability issue of mobile manipulators, particularly when the end-effector and the vehicle have to follow a predefined trajectory (for some special duties like painting a plane or carrying a light load), is a crucial subject and needs special attention. In this paper, by utilizing the manipulator compensation motions, the instantaneous proper configuration for a redundant mobile robot is determined. A fast methodology taking into account the dynamic interaction between the manipulator and the vehicle is proposed for enhancing the tipover stability (i.e. stability against overturning) of the mobile manipulator by employing the soft computing approach including a genetic algorithm, neural network, and adaptive neuro-fuzzy inference (ANFIS) controller. A genetic algorithm is utilized to find a minimum value for a function, called the performance index here, to determine the tipover stability measure. In such an algorithm, utilizing the values attained, a neural network look-up table is planned. This look-up table will expedite the operation of tracking a specified trajectory for the end-effector while the mobile manipulator simultaneously maintains its tipover stability in the most favourable manner. The benefit of using the neural network is its capacity for online control; the genetic algorithm alone cannot provide this. In this case the performance index is minimized and the rule base (the main part of an ANFIS controller) for the adaptive neuro-fuzzy controller is designed. It is considered that the tipover stability of the mobile manipulator is increased by applying the rule base of the ANFIS controller to the system. For evaluating the effectiveness and performance of the proposed algorithm, a spatial mobile manipulator is examined.