동시위치측정 및 맵핑법을 이요한 무인챠량의 장애물 회피 주행제어

Abstract

Logistics industry system is automated using an Autonomous Vehicles(AV) that is a unmanned transport device. The AV is responsible for transportation of products in the factory buildings and its use is increasing more and more. An unmanned conveying device can work at limited area or only perform limited task. Unlike the existing unmanned conveying device, Automated Guided Vehicle(AGV) is able to have the flexibility to several work environments. Nowadays, the most advanced AGV navigation method is to use laser navigation system. The laser navigation system’s scanner emits an invisible light impulse and measures the distance from AGV to all reflectors. Based on the reflectors positions, the AGV position is obtained. This method is accurate, but expensive compared with any navigation methods such as visual line, inductive guidance, magnetic tape and wall following. Furthermore, this navigation method needs several reflectors placed in the work environment. This navigation system is not suitable for unknown environment. Therefore, to solve those problems, a positioning method without any reflectors, path-generating algorithm, and control method to track the generated path autonomously are needed. This thesis presents control method using Simultaneous Localization and Mapping (SLAM) and D^*algorithms for autono- mous driving AGV with differential drive system to navigate from starting point to goal point with avoiding the stationary obstacles in industrial environment. This control method can works in unknown environment, and guarantees the reachability condition of goal point. To do this task, the followings are done. Firstly, the system configuration of AGV is described. A real AGV system is developed with several interconnected devices such as industrial PC as main controller, laser measurement system for obstacle detection, acceleration sensor and encoders for positioning, motors for actuator and batteries for power supply. Secondly, mathematic kinematic modeling of the AGV is presented to understand its characteristics and behavior. Thirdly, the SLAM algorithm based on the laser measurement system and acceleration sensor is proposed. The acceleration sensor and encoder are used for detecting the motion state of the AGV. In slippery environment and high speed AGV condition, the positioning method based on those sensors generates big error. Therefore, more advanced positioning algorithm is needed. The laser measurement sensor is not only used for detecting and mapping work environment but also proposed as the positioning sensor. Fourthly, Extended Kalman Filter is used to get the best estimation and prediction of AGV position. Extended Kalman Filter consists of prediction step and update step. In prediction step the AGV position is calculated only based on the encoder. In update step the AGV position is calculated using information from the laser measurement system. Fifthly, to go to the desired coordinate with the desired direction, a path generating algorithm is needed. D^*algorithm is proposed to generate a path from the start point to the goal point for AGV to avoid obstacles and arrive at the goal point using information of AGV, obstacles and the goal point obtained from the proposed SLAM algorithm. Using the backpoint function of the proposed D^* algorithm, a shortest path from a goal point of the AGV to a start point is generated. Sixthly, a linear controller for tracking the desired path generated by the proposed SLAM and D^*algorithm is proposed based on Lyapunov stability. To implement the proposed algorithms and the proposed linear controller, a control system is developed based on industrial PC TANK 800 and touch screen monitor. Finally, the effectiveness of the proposed algorithms and controller are verified using simulation and experiment. Several examples of experiment conditions are presented using stationary obstacles. The simulation and experimental results show that the AGV can reach the goal point successfully in all experimental conditions.

Keywords: AGV, Velocity and orientation control, Obstacle avoidance, SLAM, Laser mearurement system, D^*Algorithm, Extended Kalman Filter