Designing a time path tracking stabilizing controller for a tractor-trailer robot in the presence of uncertainties

Abstract

In this article, the discussion of track tracking in tractor-trailer systems with the condition of non-slip and non-holonomic multi-controllers is presented. One of the significant issues in the control of this system is the limitation and saturation of operators, which in real conditions causes serious challenges in the discussion of control and instability of the system. Therefore, the control of these systems, taking into account the considerations of the operator, is of great importance. The aim of this research is to design a stabilizing controller for the tractor-trailer robot in the presence of uncertainties and considering the limitations of operator saturation. In order to track the time path, the controller is presented in two kinematic and dynamic stages. The focus of this research is on the application of operator restrictions in dynamic controllers, and two control approaches are presented for that. Dynamic modeling of the robot is done using Lagrange’s equations. At first, in order to use the anti-jackknife control of the nonlinear tractor-trailer system, it is estimated by identifying the system with a linear and uncertain system. This article investigates the use of neural networks to solve kinematic problems in an omnidirectional mobile robot. The neural network is trained to learn the speed of desired angles of the robot tires, corresponding to different routes. The results show that the neural network can control the robot’s movement without dynamic and kinematic knowledge. In the next section, a neural network is used as compensation for the malfunction of the operators in a feedback linear controller in order to prevent the saturation of the operators. The weights of the neural network are updated in such a way that the stability of the system is guaranteed using the Lyapunov candidate. This article investigates the use of neural networks to solve kinematic problems in an omnidirectional mobile robot. The neural network is trained to learn the speed of desired angles of the robot tires, corresponding to different routes. The results show that the neural network can control the robot’s movement without dynamic and kinematic knowledge. Also, a comparison has been made to validate the method used in Adams and MATLAB software and the simulation conditions of the proposed method with a reference article. The simulation results show the efficiency of the designed controllers. According to the investigations carried out, it can be seen from the angular and planar error of the robot that the neural network method has a better performance than the comparative methods with an approximation between 10% and 15% and the anti-jackknife method also because of the reference method and also The anti-jack knife shows between 13% and 16% due to the complexity of the angular movement of the steering wheel and the rotation and connections of the trailer to the tractor.