Path Planning and Fault Detection for Automatic Guided Vehicle Based on Multiple Positioning Modules

Abstract

Automatic guided vehicles (AGV) increase the material handling efficiency and reduce the costs by automating manufacturing facilities or warehouse. To make the AGV move automatically, trajectory tracking control for AGV is needed. Furthermore, safety and reliability of AGV is very important factor in AGV operation. To guarantee the safety and reliability of AGV, a fault detection algorithm is required. To solve this problem, this dissertation presents an implementation and its experimental validation of path planning, trajectory tracking, positioning and fault detection algorithm for sensors and actuators of AGV system based on multiple positioning modules. To do this task, the followings are done. To design the trajectory tracking controller, mathematic modeling of the system is needed. Firstly, in this dissertation, system description and mathematical modelings of a differential drive AGV system are presented. The system description consists of mechanical design, electrical design and software design. Kinematic model of AGV is derived based on the wheel configuration and the nonholonomic constrain. Dynamic model of AGV is obtained from the Lagrangian formula. Secondly, to make the AGV move automatically, a desired path is needed. This dissertation focuses only in coverage path planning problem. This path planning algorithm can be applied to a cleaning robot, a mining robot, etc. Coverage path planning is defined as the task of finding a path that passes all points on a given area. Moreover, energy consumption also has to be considered to guarantee the optimal performance. To solve this problem, a new coverage path planning algorithm based on multi-spanning tree algorithm is proposed. In the coverage navigation of AGV, energy consumption is depends on the number of acceleration and deceleration and the total path length. The main idea to solve this problem is by reducing the turning number and the overlapped path. To do this, a minimal sum altitude algorithm (MSA) is applied to the Morse cell decomposition to minimize the turning number. Furthermore, a modification of spanning tree algorithm is introduced to minimize the overlapped path. The simulations are done to verify the superiority of the proposed algorithm compared to the existing algorithms such as vertical and horizontal cell decomposition, and conventional spanning tree algorithms. Thirdly, two trajectory tracking controllers are proposed. A first trajectory tracking controller is an adaptive backtepping controller. By choosing appropriate Lyapunov function based on its kinematic modeling of AGV with unknown slip parameter, system stability is guaranteed and a control law can be obtained. To solve this problem, an update law is proposed to estimate the unknown slip parameter. A second trajectory tracking controller is a robust servo trajectory tracking controller designed using polynomial differential operator based on internal model principle. The basic idea of the internal model principle is described. The extended system including reference and disturbance in polynomial differential equation form is introduced. The controllability checking of the extended system is done. The state feedback law is obtained by a well known regulator design method. Simulation and experimental are done to verify the effectiveness of the proposed trajectory tracking controller. Simulation and experimental results shows that the proposed controllers successfully make the AGV tracks the generated path. Fourthly, to understand the characteristics of the sensors, mathematic models of the positioning sensors are proposed. The AGV uses positioning modules such as encoder, laser scanner, and laser navigation system to obtain its position information. The basic principle and error condition of each sensor is introduced. Fifthly, a fault detection algorithm based on multiple positioning modules is proposed. The proposed fault detection method uses two or more positioning modules. In this algorithm, Extended Kalman Filter (EKF) is used to detect unexpected deviation of measurement results of two modules are affected by fault. The pairwise differences between the estimated positions obtained from sensors are called as residue. When faults occur, the residue value is greater than the threshold value. Fault identification is obtained by examining the biggest residue. Finally, to demonstrate the capability of the proposed algorithm, it is applied to a differential drive AGV system. The simulation and experimental results show that the proposed algorithm successfully detects faults when the faults occur. Finally, conclusions are presented, and the future works of the proposed algorithms are described.