Cloud Based Simultaneous Localization And Mapping For Mobile Robot

Simultaneous Localization and Mapping(SLAM) is a very important research direction for autonomous mobile robot in outdoors. Traditional robotic technologies, however, have been limited by the inherent physical constraints especially for large-scale explorations since all the computations have to be conducted in the onboard computers/microchips of the robot that have limited computing capabilities. Cloud Robotics is an emerging field of robot. It applies the cloud computing concept to the robot in order to augment the capabilities of the robot by off-loading computation and sharing the huge data or new skills via the internet. What’s more, Cloud robotics allows robots to take advantage of the rapid increase in data transfer rates to offload tasks without hard real time requirements which can improve the mobile performance of robots.In order to overcome the large amount of data, robot load equipment high and low computational efficiency problems of traditional mobile robots in SLAM, we proposes a cloud-based architecture to achieve real-time SLAM of mobile robots in outdoor environments taking advantage of the powerful computation, storage and other shared resources of the cloud. We use the architecture of C/S which treat the cloud as service and the robot as client, thus we can establish the communication between the cloud and the robots. In client, we mainly obtain data information about the mobile robot and the data information of environment. The point clouds represent the environmental information containing both the terrain points and non-terrain points. Those points need to be extracted separately from the point clouds map due to only the non-terrain points contribute to SLAM. Thus we also run the point clouds separation algorithm in client to eliminate the invalid data. The cloud as service on one hand run the ICP matching algorithm for the purpose of updating the location and the environmental information, on the other hand it in charge of the cloud data storage and processing. Due to the large amount of data, we adopt Socket programming based on TCP/IP in order to prevent data loss and other problems.The simulation experiment was carried out to ensure the accuracy and stability of the related algorithm at first. ICP matching algorithm achieved the matching and updating function of the point clouds. The point clouds separation algorithm realized the separation of the terrain point and the non-terrain point, we reduced the useless the number of points effectively for the convenience of data transmission and operation. The realization of the network transmission algorithm verified the reliability of the network data transmission. Then we finished the construction of the hardware platforms. Speed Collection Platform was used for collecting the speed information of the mobile robot, 3D Environment Platform was used for the collection of 3D point clouds environment information, Posture Information Collection Platform could get the IMU posture information of the mobile robot. We equipped the platforms on the mobile robot and selected the trees along the river in the university town as the outdoor experiment site. By analyzing the experimental data, especially the positioning error analysis and SLAM total time comparative analysis, we know that cloud robotics have great advantages in SLAM compare to the traditional SLAM methods. Cloud Robotics not only can improve the operation efficiency of the data greatly, reduce the burden of mobile robot and improve the mobile performance of robot, but also can get the online positioning operations and avoid the trouble of offline processing. At the same time we verified the accuracy of the algorithm and acquired the desired SLAM result.