Research And Implementation Of Land And Air Path Planning Algorithm Based On Multi-Sensor Fusion SLAM

As the main research technology for simultaneous positioning and map construction,multi sensor fusion SLAM is usually applied in fields,such as multi scene unmanned aerial vehicles and autonomous driving.Multi sensor SLAM technology has important application value in land and air path planning,which can improve the autonomous walking ability of unmanned aerial vehicles and the accuracy and efficiency of path planning.This research topic comes from the project cooperated by the tutor team and Beijing Institute of Technology.Through collecting environmental information using multiple sensors such as visual camera,wheel odometer and IMU,and combining it with the position and attitude information of the UAV itself,the environmental map is built in real time and completely and the location of the UAV is determined.This article establishes a grid map based on a multi-sensor fusion SLAM system,further considering the SLAM process and path planning process in land and air traffic scenarios.It improves the A* path planning algorithm in planning the optimal land air path,battery energy consumption,and operating time to achieve the comprehensive optimization of land air scene path planning.Firstly,in response to the limitations of individual sensors in the SLAM process due to their inherent characteristics,the advantages of multiple sensors are integrated for fusion positioning by analyzing the parameter model of multiple sensors.The fusion of vision-inertia-wheel odometers in land environments and vision-inertia odometer in air environments respectively forms a multi-sensor fusion SLAM system architecture,which relies on the fusion of multiple sensors to output positioning information,make the overall positioning system robust and adaptable to land and air environments.Secondly,in order to solve the problem of multi-sensor fusion odometer,a fusion algorithm for different odometers of land and air unmanned aerial vehicles in different motion states is proposed.In land motion states,a visual camera,IMU and wheeled odometer fusion algorithm based on error Kalman filtering is proposed;Propose a visual and inertial odometer fusion algorithm based on improved Kalman filtering in the state of air motion.And in order to adapt to path planning in land and air environments,a zero speed correction algorithm is used during state transition to stabilize the state of the drone.Next,an improved A* based land to air path planning algorithm is proposed in a three-dimensional environment.Under the premise of satisfying obstacle avoidance in the three-dimensional environment,a path with the best consideration in terms of time,space,and energy costs is planned,guiding the movement direction of land to air drones in the SLAM process on the ground and in the air,so that the drones can also plan the best land to air path in the land to air environment.Based on the above conclusions,conduct experimental testing and result analysis.The multi-sensor fusion SLAM system studied in this article has accurate and reliable positioning information within a certain range.The improved A* path planning algorithm based on land and air environment has significant effects,and the planned path performance parameters are better.