Comparison and Evaluation of SLAM Algorithms for AGV Navigation

The navigation technique has been always an important issue for guiding an automated guided vehicle (AGV). With the development of sensor technology, software engineering, andalgorithms, there is a spectrum of different navigation methods for AGVs. Inorder to avoid the additional environmental installation,so as to increase the flexibility of route planning, but to keep the positioning precision, more sensors, such as light detection and ranging (LIDAR), wheel encoders and gyroscope are installed on the vehicles to be automated. Some intelligent algorithms such as simultaneous localization and mapping (SLAM) algorithms and Monte Carlo localization have been developed for the navigation of vehicles, including position and orientation. The interesting question, especially for the AGVmanufacturers, is: which algorithm is more suitable for which kind of applications. The suitabilityof an algorithm for the navigation of AGVs with facilities of Light Detection and Ranging(LIDAR), encoders and gyroscope ismainly determined by four properties. They are the positioning precision, computational costs, execution time and positioning repeatability. This paper intends to investigate the suitability of an algorithm or a navigation method for AGVs with LIDAR, wheel encoders,and gyroscope. The two aspects of positioningaccuracy and repeatability are especially concerned. A general comparison of different navigation methods and algorithms is given. An experimental platform with a basic vehicle, controlling system and sensors is then developed to further evaluate the algorithms. The hardware components and software components are compatible to robot operating system (ROS). This open-source robotics middleware provides services and tools for creating robot applications. As ROS SLAM nodes, open-source SLAM algorithms can be evaluated relatively easily without any rewriting or modification of the algorithms. As a new research field, there is not jet a SLAM algorithm, which is predominant absolutely.