Implementation method of multithread distributed SLAM system based on ROS environment

Abstract

The invention discloses an implementation method of a multithread distributed SLAM system based on an ROS environment. The multithread distributed SLAM system based on the ROS environment divides thewhole state vector into a five-dimensional state including robot pose estimation and landmark estimation by use of a distributed structure, the design adopts distributed processing of a system model,multiple threads are established for multiple effective landmark points, parallel model structures forming subfilters independently are formed, then, pose estimation results of robots of the subfilters are fused in a main filter through main threads, and finally, robot position estimation results are obtained according to the fusion results of the subfilters. Finally, comparison of the algorithm in the invention with an off-line distributed algorithm through real experiments proves the feasibility and effectiveness of the algorithm, so that the multithread distributed SLAM system based on theROS environment is implemented.

Description

technical field [0001] The design of the present invention uses a distributed structure to divide the entire state vector into five-dimensional states of robot pose estimation and landmark estimation, adopts the distributed processing of the system model, and establishes multiple threads for multiple effective landmark points, forming a parallel independent The model structure of the sub-filter is formed, and then the pose estimation result of the robot of the sub-filter is fused in the main filter through the main thread, and the position estimation result of the robot is finally obtained through the fusion result of the sub-filter, so as to realize the autonomous positioning of the robot and map. Thus, a multi-sensor information data acquisition system equipped with laser, odometer, magnetic compass and other sensors and a multi-threaded distributed EKF-SLAM system prototype based on ROS environment were constructed, and the system performance was improved and optimized. B...

AI Technical Summary

Problems solved by technology

[0003] The distributed extended Kalman SLAM system based on multi-threaded online real-time processing in the ROS environment corresponds to the offline distributed SLAM system. In the offline distributed structure, sensor data acquisition and sub-filter and main filter operations are not synchronized. Yes, there is neither the problem of small amount of calculation caused by multi-threading, nor the timeliness of real-time data processing, nor does it need to consider issues such as dynamic adjustment during robot movement and clock matching between sensors

Images