Pose Detection of a Mobile Robot Based on LiDAR Data

This paper describes the application of mobile robots in the current coronavirus epidemic. Localization is a frequently discussed topic in mobile robotics research. Before a robot can start a task, it must know its current location on a map. The system proposed in this paper scans the obstacles and terrain around the robot by LiDAR to obtain a map of the environment and then uses the image recognition algorithm proposed in this paper to achieve the robot's location. This system can be applied to frontline medical robots, which can disinfect the environment or deliver medication, especially in the case of the COVID-19 epidemic, to help healthcare workers. The proposed localization algorithm is different from the traditional Adaptive Monte Carlo Localization (AMCL), which uses a 2D LiDAR sensor with image recognition to complete the localization. By using a modified template matching technique, the local map is compared with the known global map to deduce the robot's position, which is more accurate than AMCL. In this study, an indoor environment is created using Gazebo 3D environment simulation software, and a robot with a 2D LiDAR sensor is used in this environment to conduct the experiment. We designed three scenarios to validate the proposed algorithm, one with simple terrain, the second scene will appear throughout the map with other scenes of similar terrain, and the third with long straight lines. The results show that this method is feasible. © 2021 IEEE.

An Improved Localization of Mobile Robotic System Based on AMCL Algorithm

In the Mobile Robotics domain, the ability of robots to locate themselves is one of the most important events. By locating, mobile robots can obtain information about the environment and continuously track their position and direction. Among localization algorithms, the Adaptive Monte Carlo Localization (AMCL) algorithm is applied most often in robot localization, a two-dimensional environment probabilistic localization system to improve the problems such as high computational complexity and hijacking of mobile robots that exist in the traditional MCL method. The proposed method is based on 2D laser information, range finder information, and AMCL to accomplish the localization task. Furthermore, an optimized AMCL algorithm is proposed to increase the accuracy of localization in terrain that is easy to fail to locate, have a chance to locate successfully when a localization error occurs, and apply the optimized AMCL to the mobile robot system. From the experimental results, we know that the improved AMCL algorithm can enhance the positioning accuracy of the robot effectively, which has better practicality than the original AMCL.