Research On Adaptive Active Navigation Technology Based On Multi-sensor Fusion

With the increasing application of robot technology in modern industries,people’s requirements for robot intelligence are getting higher and higher.At present,most robots need to provide prior map information to achieve adaptive active navigation.When in an unknown environment,the robot’s navigation function is greatly limited.Real-time localization mapping and adaptive active navigation based on multi-sensor fusion have become the key technologies to remove such limitations,and have been paid more and more attention by more and more researchers and developers.In this paper,a robot adaptive active navigation system and a corresponding experimental platform are built on the basis of four sensor data fusion of single-line lidar,depth vision sensor,wheel encoder and attitude sensor.In addtion,the multi-sensor data fusion method,the navigation path planning and the optimization of the target point active selection strategy are studied,so as to realize the adaptive active navigation function in the unknown environment of the robot experimental platform.The main research contents are as follows:At first,the target demand is analyzed according to the requirements of the robot navigation system.By establishing the target function of the adaptive active navigation system and building the overall framework of the system,the robot experimental platform equipment selection and software modular design ideas are explained.Second,the multi-sensor fusion real-time localization mapping.is completed.According to the proposed environmental information fusion method,the depth vision sensor data is converted into single-line lidar virtual data,and the virtual data and real data are fused,which improves the accuracy of the system to construct a two-dimensional grid map.At the same time,according to the fusion method of odometer data and scanning matching positioning information,the wheel encoder,attitude sensor data and scanning matching positioning information are fused to improve the accuracy of odometer data and eliminate accumulated errors.Then,the robot adaptive active navigation is realized.According to the optimized A-Star search algorithm and artificial potential field method,the navigation path planning and local obstacle avoidance functions are completed,and the robot adaptive navigation function is realized.At the same time,based on the fast expanding random tree algorithm,the optimal navigation target point selection strategy which enables the robot to complete operations such as boundary point generation,boundary point evaluation,firefly optimization,and autonomous selection of the best target point in a known local map is proposed,so as to achieve the purpose of active robot navigation.At last,through experiments to verify the system function and analyze the results.The robot experiment platform is placed in a real unknown scene in advance.Under the condition of no human control,the robot experiment platform independently will select the best navigation target point and plans the navigation path.At the same time,the environmental map information is constructed.Through the analysis of results to verify the practicability and robustness of the adaptive active navigation system based on multi-sensor fusion for navigation tasks in unknown closed environments.