Research Of Mobile Robot Localization Based On Multi-sensor

Intelligent mobile robots are an important research field for the new generation of artificial intelligence.Demands in many fields such as factories,medical treatment and logistics are increasing.During the epidemic,a large number of intelligent mobile robots rushed to the front line of the epidemic,fighting side by side with medical staff in the areas of drug distribution,ward disinfection,security patrols,and security checks,and played an important role.Among the key technologies of mobile robots,stable,efficient and accurate localization is the core to achieve autonomy.However,the actual environment is complex and diverse,and the position in the movement process is easy to lose,and a single sensor has its own limitations.Fusion of multiple sensors to improve the localization ability will promote the application of mobile robots in actual scenes.Therefore,aiming at the localization problem of mobile robots,this paper studies the localization algorithm based on multiple sensors to improve the relocation and position loss recovery capabilities.The main research contents and results include:(1)The advantages and disadvantages of common sensors such as wheel odometer,inertial measurement unit(IMU),camera and Li DAR in intelligent mobile robot system are analyzed,and the significance of multi-sensor fusion is expounded.The typical single sensor navigation and localization algorithm is studied,and the key technologies of vision and Li DAR localization algorithms are emphatically explored,and their mathematical models are derived.The localization algorithm based on Li DAR and the visual localization algorithm based on the feature point method are implemented respectively,and the localization effect and respective failure conditions are verified by designing experiments.(2)In order to solve the systematic error of the wheel odometer,a least square problem with the Li DAR scan matching as the real value and the wheel odometer data as the observed value was constructed to solve and calibrate the wheel odometer.Aiming at the random error of the wheel odometer,the extended Kalman filter method was used to fuse the IMU data and the wheel odometer data,and the fusion results were used as the initial value of the Li DAR scan matching to estimate the local location.(3)The problem of starting at any position and losing the navigation position is studied,and the mode conversion module is designed to switch according to the motion state.During normal navigation,occupancy grid map by Li DAR is beneficial to path planning,and when the position is lost,the visual system with rich information is used to relocate the position.Through experiments,it is found that different locations on the map have different help for localization.Therefore,a recovery point strategy is designed.The visual feature points are clustered,the cluster center is used as the recovery point,and the mileage information is saved.When the position is lost,the mileage in the continuous pose is used.The data is used as the pose estimation value after the position is lost,and it moves directionally toward the recovery point,so that the localization algorithm has a stronger recovery ability after the position is lost,and improves the stability of the operation in a dynamic environment.(4)The simulation environment built by Gazebo software is configured according to the attributes of the physical robot.The debugging algorithm is efficient and cost saving.In order to verify the performance of the algorithm in the real environment,a physical robot platform was built according to the performance and structure requirements.Through the design experiment,it is verified that the designed localization algorithm has good results in localization accuracy,localization time,correction of the wrong position,localization success rate and recovery ability after the loss of pose.In conclusion,this article in view of the mobile robot localization problem,combined with multiple sensor was designed which has the function of mode conversion localization algorithm,designed the map save and load module,propose a location recovery strategy,set up the experimental platform experiment to verify the effectiveness and practicality of the system,and will promote the application of mobile robot in the actual scene.