Research And Implementation Of SLAM Mapping Algorithm For Mobile Robot Based On Multi-Sensor

In the field of robot research,it is more and more important to make mobile robot have a higher level of intelligence in the future practical life.Slam technology is the foundation of intelligent mobile robot in the future.Using multiple sensors for slam is the future development trend.For this reason,this project is devoted to the research of using multi-sensor to build grid map for autonomous path planning,and improving the vision slam system to build three-dimensional dense point cloud map,which lays the foundation for autonomous navigation of robot in dense map.Build Multi-sensor Detection System.The traditional two-dimensional lidar and rgb-d sensor are used as the main sensors of the mobile robot to construct the map.The two sensors collect the surrounding environment data at the same time,use the data collected at the same time to construct the grid map of the environment using the gmapping-slam algorithm framework,and then use the Bayesian probability model for data fusion to get the final results.In the simulation environment of gazebo,the simulation experiments show that the multi-sensor detection system is more accurate and informative than the tradi tional grid map which only uses lidar,and is more suitable for autonomous path planning of mobile robots.It lays a foundation for further adding other sensor data to build more accurate map.Build a 3D dense point cloud map of the environment.In this pa per,we add a thread to build dense point cloud map,so that the algorithm framework can build dense point cloud map on the hardware platform in real time.Using the data set of tums to verify the orb feature points used in this paper and the SLAM algorith m proposed in this paper,the results verify the effectiveness of this algorithm.The improved slam system provides a precondition for mobile robots to achieve autonomous navigation in unfamiliar environment.Experimental verification is carried out in the simulation environment and the actual scene.The working mechanism and the core concepts of the robot operating system,which is the software framework used in the experiment,are introduced briefly.Using this distributed software framework can greatly i mprove the computing speed.The traditional feature points and the orb feature points used in this paper are extracted on the open data set.Through comparison,it is found that the orb feature points extraction speed is the fastest,the distribution is th e most uniform,and it has the ability to build dense map in real time.The experiment environment is built to let the mobile robot experiment platform run the slam system in the built environment,and compared with the traditional algorithm,it is proved that the algorithm in this paper has better loop detection characteristics,and the map accuracy is higher.